Next Article in Journal
Measuring the Social Perception of Religious Freedom: A Sociological Perspective
Previous Article in Journal
On the Origin and Conceptual Development of ‘Essence-Function’ (ti-yong)
Previous Article in Special Issue
The King Must Protect the Difference: The Juridical Foundations of Tantric Knowledge
Article Menu
Issue 4 (April) cover image

Export Article

Religions 2019, 10(4), 273;

Kālavañcana in the Konkan: How a Vajrayāna Haṭhayoga Tradition Cheated Buddhism’s Death in India
School of Languages, Cultures and Linguistics, SOAS University of London, Thornhaugh Street, Russell Square, London WC1H 0XG, UK
Received: 26 February 2019 / Accepted: 12 March 2019 / Published: 16 April 2019


In recent decades the relationship between tantric traditions of Buddhism and Śaivism has been the subject of sustained scholarly enquiry. This article looks at a specific aspect of this relationship, that between Buddhist and Śaiva traditions of practitioners of physical yoga, which came to be categorised in Sanskrit texts as haṭhayoga. Taking as its starting point the recent identification as Buddhist of the c.11th-century Amṛtasiddhi, which is the earliest text to teach any of the methods of haṭhayoga and whose teachings are found in many subsequent non-Buddhist works, the article draws on a range of textual and material sources to identify the Konkan site of Kadri as a key location for the transition from Buddhist to Nāth Śaiva haṭhayoga traditions, and proposes that this transition may provide a model for how Buddhist teachings survived elsewhere in India after Buddhism’s demise there as a formal religion.
Śaivism; Buddhism; Hinduism; Tantra; Yoga; India

1. Introduction

It has long been recognised by indologists that Vajrayāna Buddhist and Nāth1 Śaiva traditions have much in common, in particular adepts, sacred sites and metaphysical terminology. In recent years scholars have explained these commonalities either by pointing to the Nāths as their originators or by claiming that the two traditions share a common substratum. Early 20th-century Indian scholars, on the other hand, viewed Vajrayāna Buddhism as their source.2
The relationship between Vajrayāna and Nāth Śaiva traditions is but one part of the complex relationship between Vajrayāna and Śaivism as a whole, which has been the subject of detailed analysis since an article by Alexis Sanderson published in 1994 in which he demonstrated the dependence of certain Buddhist Yoginītantras on texts of the Śaiva Vidyāpīṭha.3 Shaman Hatley has shown that since the time of the earliest tantric texts there are likely to have been borrowings between the two traditions in both directions, albeit on a smaller scale than in the examples provided by Sanderson.4 David Seyfort Ruegg and Francesco Sferra, while accepting the borrowings demonstrated by Sanderson, have argued the case for a shared substrate, with Sferra proposing that “Buddhist and Hindu Tantric traditions only appear to be distant from one another at the theoretical level when the common practices and ‘substratum’ are imbued with a doctrinal content”.5 Sanderson has rejected the concept of a shared substratum because it is an entity that is only inferred, whereas everything we perceive in this context is either Śaiva or Buddhist.6
Recent studies in the field of haṭhayoga, which in previous scholarship is usually said to have originated within Śaiva Nāth traditions, have shown that the term haṭhayoga is first found in a range of Vajrayāna works dating from the 8th century onwards7 and that the c. 11th-century Amṛtasiddhi, the first text to teach any of haṭhayoga ’s distinctive practices and principles, was composed in a Vajrayāna milieu.8 The Amṛtasiddhi was directly drawn upon in the production of several subsequent haṭha texts, none of which was Buddhist and the earliest of which was probably the Nāth Śaiva Amaraughaprabodha.9 In this paper I shall draw upon textual, epigraphic and material sources to show that the Konkan (the coastal region of modern-day Maharashtra, Goa and Karnataka), and in particular the Kadri monastery in Mangalore, is likely to be where the Amaraughaprabodha was composed, and that its composition was symptomatic of the appropriation from Vajrayāna Buddhists by the Śaiva Nāths of not only practice and terminology, but also the Kadri monastery itself. Unlike the parallels between earlier Vajrayāna and Śaiva traditions that have been the focus of recent scholarship, this transition coincides with the demise of Buddhism in India. The evidence presented suggests that the Kadri Vajrayāna Buddhist tradition was not destroyed or expelled, but absorbed into that of the Nāths, and provides a possible model for such transitions elsewhere.

2. Virūpākṣa

The Amṛtasiddhi makes no mention of any of the more celebrated adepts shared by the Vajrayāna and Nāth traditions, such as Matsyendra or Gorakṣa. Instead it attributes its teachings to a renowned Vajrayāna siddha called Virūpākṣa (or sometimes Virūpa), who is little known outside of Tibetan Buddhist traditions, in which he first appears in perhaps the 12th century.10 Tibetan hagiographic treatments of Virūpākṣa and textual cycles associated with him are particularly rich and diverse, 11 as are his depictions in Tibetan paintings12 and statuary. Here I shall draw upon these Tibetan materials to note only (1) that they indicate that after spending his early life in east India, Virūpākṣa was active in the Deccan and the south;13 (2) that the Sanskrit Amṛtasiddhi, whose teachings were attributed to Virūpākṣa, was translated into Tibetan (probably in the late 11th century)14 and an associated cycle of Tibetan texts (usually referred to in Sanskrit back-translation as Amarasiddhi),15 whose teachings were also attributed to Virūpākṣa, developed soon after; and (3) that Virūpākṣa is said to be the human conduit of the Tibetan Lamdre (lam ’bras) tradition, whose teachings include the practices of Trulkhor (’krul ’khor), some of which correspond to the three central techniques of the Amṛtasiddhi and which are predicated on an understanding of the yogic body first taught in the Amṛtasiddhi.16
The siddha Virūpākṣa has left few traces in Indian material and textual sources. Buddhist hagiographies mention a shrine to him at Somnath,17 but there is no evidence of one there now. The temple of Bhīmeśvara at Draksharama in Telangana, which predates 1130 ce18 and is associated with the Virūpākṣa legend in early Marathi texts and current Tibetan hagiographies, has a shrine to him; there are reliefs depicting him on the c. 1230 ce Mahudi Gate at Dabhoi in Gujarat (Figure 1), on the exterior wall of a c. 13th-century cave at Panhale Kaji in the north Konkan (Figure 2) and perhaps on the c. 14th-century Someśvara temple at Pimpri Dumala near Pune;19 there is a Virūpākṣa cave on Arunachala hill in which Ramaṇa Maharṣi spent long periods in retreat; and the Kāmākhyā temple in Assam has a relief which depicts him.20
Only one Indian Buddhist text other than the Amṛtasiddhi mentions Virūpākṣa: the Caryāgīti, a c. 11th-century collection of fifty middle Indic dohā verses attributed to various siddhas. One of its dohās is by Biruā (i.e., Virūpa/Virūpākṣa), who, in highly esoteric language, summarises a yoga method which is similar to that of the Amṛtasiddhi but is couched in a metaphor of alcohol production rather than the Amṛtasiddhi’s alchemy.21 The Caryāgīti’s place of composition is uncertain.22 I know of eight non-Buddhist texts which mention Virūpākṣa. One is the Varṇaratnākara, an early 14th-century Maithili compendium on a variety of subjects which gives a list of 84 siddhas and includes the name Virūpa.23 The remaining seven texts are connected in some way with the Śaiva Nāth tradition and are from the Deccan or south India. The two oldest are Mahānubhava works in Marathi (which corroborates statements elsewhere that Virūpākṣa was from the Maratha region):24 the c. 1280 ce Līḷācaritra, a hagiography of the Mahānubhava guru Cakradhara composed by his devotees,25 and the Tattvasāra of Cāṅgadeva, a compendium of Mahānubhava teachings which was completed in 1312 ce. After these Marathi works, the next texts to mention Virūpākṣa are the c. 1400 ce Telugu Navanāthacaritramu, an account of the lives of the nine Nāths composed by Gauraṇa at Srisailam, the contemporaneous Vikramārkacaritramu of Jakkana, which includes Virūpākṣa’s name among those of the nine Nāths,26 and the Sanskrit Haṭhapradīpikā, a compilation by Svātmārāma of teachings on haṭhayoga whose parallels with the Navanāthacaritramu (and other Telugu texts) suggest it is from the same period and region. The Tārārahasya, a 16th-century Sanskrit treatise on the worship of the goddess Tārā by the Bengali author Brahmānandagiri, includes Virūpākṣa in a list of eight human gurus to be worshipped.27 The last non-Buddhist text to mention Virūpākṣa is the c. 17th-century Sanskrit Kadalīmañjunāthamāhātmya, a celebration of the temple of Mañjunātha at Kadalī (now known as Kadri, a part of the coastal town of Mangalore), in which Virūpākṣa is again one of nine Nāths, seven of whose stories, including that of Virūpākṣa, are taught in extenso.
In the Līḷācaritra, Cakradhara tells his disciples how Virūpākṣa broke in two the Bhīmeśvara śivaliṅga at Drākṣārāma in modern-day Telangana. Gorakṣa converted Virūpākṣa from vajraolī, i.e., the [Buddhist] Vajra lineage, to amaraolī, the [Śaiva] Amara lineage, and gave him the name Adaṇḍīnātha.28 The Marathi Tattvasāra includes Virūpākṣa in a list of 84 siddhas.29
In the Navanāthacaritramu, Virūpākṣa is one of the nine Nāths of the text’s title and his life story is told in detail. He is the second son of king Gaṇyāvanta and Añjani, who were from Maharashtra.30 Matsyendranātha, the first of the nine Nāths and the guru of the other eight, meets him in a forest. After Virūpākṣa recounts how he has eaten the heart of a bird which a hunter had informed him would make him a siddha,31 a voice from the sky confirms his story and tells that of his previous birth, at which Matsyendra initiates him and instructs Gorakṣa to teach him yoga. After receiving instruction, Virūpākṣa travels to the Karṇāṭa, Lalita, Kanauj, and Māḷava regions, and initiates several disciples.32 In his Vikramārkacaritramu, Jakkana gives a list of nine Nāths almost identical to that of Gauraṇa. He does not tell their stories but for each of them highlights one characteristic, which for Virūpākṣa is devotion.33
In the Haṭhapradīpikā nothing is said about Virūpākṣa other than that he, along with 28 other mahāsiddhas, broke the rod of time by means of the power of haṭhayoga.34 Virūpākṣa is at number 11 in the list, immediately after Cauraṅgi, Mīna (who is here differentiated from Matsyendra)35 and Gorakṣa. As will be explored in more detail below, these three siddhas are closely associated with Kadri and the Amaraughaprabodha,36 approximately twenty verses from which are found in the Haṭhapradīpikā.
In six chapters and more than 300 verses the Kadalīmañjunāthamāhātmya gives a detailed account of several episodes in Virūpākṣa’s life.37 He is born to a pious Brahmin couple from the northern Konkan as an incarnation of Mañjunātha of Kadalīvana, i.e., the Śiva of the Kadri temple and monastery complex. He wanders across India and has various adventures, including the conquest of a demon threatening all the gods; the assumption of the appearance of an ascetic and subsequent humbling of the Veda-obsessed brahmins of Drākṣārāma (an episode which brings to mind the Līḷācaritra’s story of his breaking of the Bhīmeśvara liṅga at Drākṣārāma);38 and a sojourn in Kāñcī, where he teaches the citizens by day and sports with women by night. This last episode includes an echo of the popular Tibetan story in which Virūpākṣa stops the sun in its course until the king pays the bill for his drinks.39 Here he falls for one of the women of Kāñcī and, in order to impress her, grabs the moon and makes it into a goblet with which to ply her with drink. The gods become concerned and, at Brahmā’s instruction, go to Virūpākṣa and sing his praises, at which he puts the moon back in its rightful place.
With this, the historical trail left in India by Virūpākṣa goes cold,40 but he has left enough clues for a tentative identification of the region in which his teachings were transmitted from Vajrayāna Buddhism to Nāth Śaivism. As noted above, Tibetan hagiographies point to south India and the Deccan as being central to his activities and almost all Indian material and textual sources associated with him are from the Konkan and Deccan.41 The Navanāthacaritramu identifies him as the author of the Amṛtasiddhi42 and anchors his story in the Deccan: the Navanāthacaritramu was composed at the request of the pontiff of the Bhikṣāvṛtti monastery in Srisailam43 and Virūpākṣa is said therein to have been born in present-day Maharashtra and to have travelled to Karnataka.44 The Haṭhapradīpikā adds nothing to his lore, but strengthens the connection with Srisailam since it has many parallels with the Navanāthacaritramu and was probably composed in the same region and in the same period.45 The Kadalīmañjunāthamāhātmya, one of whose manuscripts is likely to have been copied in 1730 ce and may not, in its present form, be much older than that, but which preserves some old legends from the Kadri site,46 identifies Virūpākṣa with Śiva Mañjunātha of Kadri. All of the Nāths whose stories are told in the Kadalīmañjunāthamāhātmya are similarly identified with Mañjunātha, but Virūpākṣa is further tied to the region by virtue of being the only one to be given the epithets koṅkaṇādhīśa, koṅkaṇeśvara and koṅkaṇa.47

3. Nāth Śaivism and Vajrayāna in the Konkan

Virūpākṣa’s legend thus points to the south of the Indian subcontinent and in particular the Konkan as the likely location of the transition of his teachings from Vajrayāna to Nāth Śaivism, but gives little detail of how it might have happened, with only the Marathi Līḷācaritra actually indicating a transition from Buddhism to Śaivism. I shall now widen the enquiry from Virūpākṣa to evidence for the presence of Vajrayāna and Nāth Śaivism in the Konkan, which will further support the supposition that such a transition occurred there and provide more detail on how it happened.

3.1. Nāth Śaivism in the Konkan

Nāth Śaivism has a long history in the Konkan. Western India and the Konkan region were important centres of early Paścimāmnāya Kaula Śaivism, elements of which were preserved by subsequent Nāth lineages.48 Matsyendranātha, the revealer of the Kaula doctrine in the Kali age and the first human guru of the later Nāth sampradāya, is closely associated with the region. His consort was called Kuṅkuṇāmbā, “the mother of the Konkan”;49 a c. 13th-century anthology of his teachings, the Matsyendrasaṃhitā, is associated with the Konkan and was composed either there or in the Tamil region;50 and among the earliest material evidence of the Nāth sampradāya is the c. 13th-century statuary at Panhale Kaji on the Konkan coast between Mumbai and Goa, which includes at least two images of Matsyendra.51
The monastery at Kadri is today the most important Nāth site in south India. Statues of three Nāths at the Mañjunātha temple below the monastery have been dated on stylistic grounds to the 14th or 15th centuries.52 The first firmly datable evidence of the presence of the Nāths at Kadri is in a Kannada inscription of 1475 which records a grant of land to Maṅgalanātha Oḍeya “at the pure place of Śiva, Kadire”, the income from which will support the worship of Gōrakṣanātha and Candranātha, a deity at the temple.53 Bhatt notes an inscription from Bārakūru dated 1490 ce which refers to the position of arasu or king at Kadri in the context of one Subuddhinātha Oḍeya, disciple of Anupamanātha Oḍeya, and to the worship of Gōrakhanātha at the maṭha of the latter.54 Subsequent descriptions of the head of the Kadri monastery also call him the “King of the Yogīs” and until the demise of the Vijayanagara kingdom and resultant depredations by the Nāyaka kings he lived in great style. Writing at the beginning of the 16th century, the Italian traveller Ludovico di Varthema said that the king ruled over 30,000 people and travelled about India with an impressive entourage including a troop of warrior yogis.55 By the 17th century the king lived in much reduced circumstances56 and it seems that at this point the Kadri monastery was taken over by a northern Nāth lineage intent on creating a pan-Indian yogi order. By 1820 ce Kadri was recognised as the southern seat of the four seats of the Nāth sampradāya.57 To this day the head of the Kadri monastery, who retains the title of “King Yogi” (rājā yogī), is drawn from northern Nāth lineages. The changes at Kadri in the 17th century resulted in the migration southwards from Mangalore to northern Kerala of the yogi caste associated with the previous royal lineage.58 Some members of the caste remained at Kadri, but they have a fraught relationship with the King Yogi, whom they look to for leadership but who cannot speak their language. The King Yogi is elected once every 12 years at a council of the Nāth sampradāya at the Nasik Simhastha festival (which nowadays is recognised as one of the four Kumbh Melas), after which he and several hundred Nāths walk barefoot for six months to Kadri, where he is installed in a lengthy consecration ritual called the rājyābhiṣeka, which has parallels with traditional royal initiations.59

3.2. Vajrayāna in the Konkan

Buddhism was well established in the Konkan when Hsüan-Tsang visited the region in the seventh century,60 and Vajrayāna had a small but significant presence there and in surrounding areas from the sixth to thirteenth centuries, and perhaps later. Sixth-century statues of Tārā and Avalokiteśvara are found in the western Deccan.61 The colophon of the ninth-century Cakrasaṃvarapañjikā of Jayabhadra (who was also known as Koṅkaṇapāda) says that its author visited a temple of Tārā at the Konkan site of Mahābimba.62 A statue of Mañjughoṣa from the Kadri monastery but now in the Mangalore government museum dates to the ninth century or earlier.63 One of the 29 caves at Panhale Kaji contains a tenth-century image of the Vajrayāna deity Acala (Figure 3).
The early ninth-century Vajrayāna adepts Dharmākara and his fellow initiate Pālitapāda lived in the Konkan.64 Pālitapāda was twice visited there, probably at Kadri, by Jñānapāda, the founder of an important eponymous tradition of exegesis of the Guhyasamājatantra.65 The Vajrayāna teacher Śākyamitra visited the Konkan region in the tenth century66 and the same century saw the composition of the Hevajratantra, which includes Koṅkaṇa in a list of 24 sacred sites.67 An illustrated manuscript of the Aṣṭasāhasrikā Prajñāpāramitā dated 1015 ce68 includes six sites in the Konkan among 72 Vajrayāna places of worship.69 An inscription at Baḷḷigāve in Karnataka (80 kilometres inland from Gokarṇa) dated 1065 ce records the construction of a vihāra containing images of Buddha, Lokeśvara and Tārā Bhagavatī70 and another inscription from the same location dated 1067 ce records the establishment of a temple of Tārā Bhagavatī.71 An 11th-century stone statue of Tārā from Baḷḷigāve is still visible at the site and may be one of those mentioned in the contemporaneous inscriptions.72 As will be explored in more detail below, the Kadri Mañjunātha temple contains bronze sculptures of the Vajrayāna deities Lokeśvara and Mañjuvajra (as well as a bronze of the Buddha), with an 11th-century inscription recording the establishment of the Lokeśvara image in the kadirikā vihāra, i.e., a Buddhist monastery at Kadri. An image dated to the 11th century in the Lokanātha temple at Haṭṭiyaṅgaḍi, Coondapur, might be of Jāmbāla.73 A 12th-century inscription at Dharmavolal (today known as Ḍambaḷ, 60 km east of Hubli) records the worship of Buddha and Tārā.74 An image of Akṣobhya from Puttige, Mūḍabidure, dates to the 12th or 13th century.75 A 13th-century inscription from the village of Koḷivaḍ (20 km east of Hubli) records the worship of Tārā.76 According to the Jain exegete Vīrānandī, Buddhist ascetics called Ājīvakas were active in the Kanara region in the 12th century.77 Finally, two Tuḷu inscriptions, one dating to 1187 ce, the other to 1545 ce, mention Buddhism.78
The accounts of tantric Buddhism in India by the 17th-century Tibetan scholar Tāranātha include several references to the Konkan. Tāranātha’s histories are notoriously unreliable, but some of his reports of teachers visiting the Konkan are corroborated by the older evidence noted above. Thus he says that Jayabhadra lived for some time at Mahābimba,79 describes Jñānapāda’s visit to Pālitapāda and his initiation into the Guhyasamājatantra,80 and confirms that Pālitapāda’s co-initiate Dharmākara was from the Konkan.81 Tāranātha makes several more mentions of the Konkan in his accounts of early Vajrayāna adepts and teachers.82
Writing of more recent events, Tāranātha says that Śāntigupta, the guru of his own guru Buddhaguptanātha, who lived during the 16th century and had close links with the Nāth saṃpradāya, went to Suvarṇadhvaja in the Konkan, which was “...a noble well-proportioned place. Its monastic colleges flourished. There were about fifty fully ordained monks there and at most about one thousand upasakas”.83 Subsequently “[t]he monks of Maharata and Kongkuna invited him and he went to all their temples giving empowerments, upadesas, alms, sermons on the tantras, etc., and he clearly explained the Vajrayana teachings”.84
These references to a flourishing Vajrayāna tradition in 16th-century Konkan are intriguing, but are likely to be garbled reports from earlier times. In Tāranātha’s detailed account of the travels of his guru Buddhaguptanātha, he writes the following:
Then in Koṅkaṇa he embarked and went to the west up to an island called ḥgro ling[,] in Sanskrit Dramiladvīpa. In the language of the Muhammadans, the barbarians and [the inhabitants] of the small island, it is called la sam lo ra na so (in Śambh: sam lo ra na so). In that island the teachings of the guhyamantras are largely diffused. He heard these from a paṇḍit called Sumati who had acquired the mystic revelations (abhijñā), the mystic power of the Saṃvara (tantra) and of the Hevajra (tantra) and then he learnt the detailed explanation of the Hevajratantra. This Hevajratantra belongs to the system of the Ācārya Padmasambhava. Generally speaking, the tradition of the fourfold tantras is still uninterrupted in that island, and if we except the sublime and largely diffused Kālacakratantra, whatever is in India is also there such as the (Vajra)kīlatantra and the Tantra of the daśakrodhas, many Heruka-tantras, Vajrapāṇi, mkhaḥ ldiṅ (Garuḍa), Māmakī, Mahākāla, etc. Then the sublime order of Hayagrīva which is largely spread in India is to be found there. Moreover there are many sacred teachings (chos) belonging to the Tantras expounded by Padmasambhava. Though the community is numerous, the rules of the discipline are not so pure. The monks wear black garments and usually drink intoxicating liquors …”.
(Tucci 1931, p. 690)
The likely location for this small island is among those near Karwar on the coast of northern Karnataka, with Anjediva the best candidate,85 but it and all the other islands off the Konkan coast were taken over by the Portuguese at the beginning of the 16th century, well before Buddhaguptanātha would have visited the region.86 Tāranātha appears to have been mistaking Jesuit priests for bibulous black-clad Buddhists.87

4. Vajrayāna-Śaiva Interaction

Of the Konkan sites mentioned above, there are two where both Vajrayāna and Śaiva traditions flourished and which are thus possible locations for a transition from the former to the latter.88

4.1. Panhale Kaji

At Panhale Kaji, which is eight kilometres inland from the Konkan coast about halfway between Mumbai and Goa, there are 29 rock-cut caves dating from approximately the 6th century ce onwards.89 As noted above, one of them houses a statue of the Vajrayāna deity Acala. Two others contain multiple images of Nāths, including depictions of Matsyendra overhearing Śiva teach Pārvatī the Kaula doctrine, Virūpākṣa,90 groups of 9 (or perhaps 10) and 84 siddhas, and a relief of Tripurasundarī. An impressive but damaged statue of Gorakṣa was also found at the site when it was restored in the 1970s.91
There are no textual or epigraphic sources to suggest that either the Vajrayāna or Nāth presence at Panhale Kaji was of wider significance, and the site fell into disuse from the 14th century onwards.

4.2. Kadri

It is the Kadri Mañjunātha temple and monastery complex that gives us our best evidence of links between Vajrayāna and Nāth traditions in the Konkan, with a wealth of material showing that both traditions flourished there.92 There is a gap of more than four centuries between the last firmly datable evidence of Vajrayāna Buddhism at the site and the first of a Nāth presence, but there is much to indicate continuity between the two.
The Kadri statuary is a key indicator of such continuity. The three bronzes housed in passages to the north and south of the central shrine of Mañjunātha (a svayambhū or “self-born” liṅga representing Śiva), are perhaps the finest Buddhist images still worshipped in India (albeit now as Hindu deities, on which see below). The inscription on the plinth of the single bronze in the corridor on the south side of the Mañjunātha shrine records how the Āḷupa king Kundavarman established an image of the god Lokeśvara in the Kadirikā vihāra on the 16th of January 1068 ce.93 The iconography and workmanship of the image support this date and the two other bronzes appear to be of a similar age.
The three images are currently worshipped as Brahmā, Viṣṇu and Vyāsa, and have been since at least 1730 ce because they are identified as such in the Kadalīmañjunāthamāhātmya.94 Their iconography, however, is clearly Buddhist. The statue on the inscribed plinth is worshipped as Brahmā, but is in fact a form of the Bodhisattva Mañjuśrī, most likely Mañjuvajra (Figure 4).95
Of the two bronzes in the northern corridor, that which is worshipped as Viṣṇu is a form of the Bodhisattva Lokeśvara (better known as Avalokiteśvara), who can be identified by the representation of the cosmic Buddha Amitābha on his crown of matted hair, his antelope skin, his hand gestures and his sattvaparyaṅka seated position (Figure 5).96
The bronze worshipped as Vyāsa is the Buddha, in a form which closely matches Sri Lankan, Tamil and some Southeast Asian Buddha images; his seated position with the legs half-crossed (sattvaparyaṅka) rarely occurs in Buddha images from India, apart from in the Tamil region (Figure 6).97
These identifications are well supported by comparison with other such images. Muddying the waters, however, is the inscription on the plinth of the Mañjuvajra image, which records the establishment of an image of Lokeśvara. It seems that at some point the image of Lokeśvara was removed from its original plinth and replaced with the image of Mañjuvajra.98
In the Mangalore Museum is a stone statue of Matsyendranātha, the first human guru of the Nāth saṃpradāya (Figure 7).99 The statue is said in its label to be from the Kadri monastery and is cracked across its neck, which may account for its having been removed from its original location.
The mūrti shares many features with the Lokeśvara bronze: it is three and a half feet high; it is seated in sattvaparyaṅka, an easy cross-legged position; its right hand rests on its right thigh, palm open;100 and it has a crown of jaṭā, matted locks, on which is depicted an indistinct icon. This latter is of particular significance, because no other non-Buddhist sculpture in India includes such an element.
Today the Matsyendra shrine in the Kadri monastery houses an apparently newer and whole mūrti of Matsyendra, which is very similar to that found in the museum and was presumably made to replace it (Figure 8).
There are two small but significant differences between the two Matsyendra images. In both statues Matsyendra wears on a thread around his neck a siṅgī, the small horn signifying membership of a Nāth lineage. That on the older statue is shaped like an antelope’s horn, while that on the later statue is a whistle like that worn by yogis of today’s Nāth saṃpradāya. Mughal miniatures depict the earlier style of horn from the mid 16th-century onwards; the whistles are not seen until the 18th century.101 On its label in the museum, the older image is dated to the 10th century. I do not know the grounds for this dating, but Bhatt (1975, plate 302(b)) gives the same date so may be its source. He accepts the incorrect reading gate found in the published transcription of the inscription on the plinth of “Lokeśvara” (i.e., Mañjuvajra) so takes the image to be dated 968 ce and proposes the 10th-century date for Matsyendra because he believes the Nāth tradition to have predated Buddhism at the site. I see no need to propose such an early date for the Matsyendra image in the Mangalore Museum. The other difference between the two images of Matsyendra is that the icon on the crown of jaṭā on the older image is unclear (Figure 9), unlike that of the newer one (Figure 10).
Images of Bodhisattvas wear small representations of one of the five Tathāgatas, indicating their lineage. The cosmic Buddha on Lokeśvara is Amitābha (Figure 11) and that on Mañjuvajra (and on his prabhāvalī) is Akṣobhya (Figure 12). The icon on the newer Matsyendra is clearly a four-armed deity with the jaṭāmukuṭa, club and fish seat of the Matsyendra image itself, together with the ḍamaru drum usually carried by Śiva, especially in his Bhairava form. Pai (1946, p. 64) identifies this icon as Ādinātha, i.e., the form of Śiva seen as the founder of the Nāth saṃpradāya. The fish seat, however, which is also faintly identifiable on the older Matsyendra image in the Mangalore Museum, indicates that this is an image of Matsyendra himself, while his four arms suggest that he has been deified.
The 1068 ce inscription on the plinth supporting the image of Mañjughoṣa at the Mañjunātha temple says that an image of Lokeśvara was established in the Kadirikā vihāra, i.e., a Buddhist monastery at Kadri. No mention of a separate temple is made in the inscription, but today there are two sites at Kadri, the Mañjunātha temple and, a few minutes walk up a neighbouring hill, the Kadri maṭha or monastery. It is unclear whether the Kadri temple and monastery have always been separate sites. The first reference to the existence of both is from the report of a visit to Mangalore of Pietro della Valle in 1624 ce.102 There are meditation caves at the monastery which date to approximately the 12th to 13th centuries ce.103 The oldest inscription in the temple courtyard is dated śaka 1308, i.e., 1385 or 1386 ce.104 Today the temple is in the control of Mādhva priests and the Nāth presence is confined to the monastery,105 but three stone statues in the middle of the south, west and north exterior walls of the inner shrine of the temple which sport yogi iconography (earrings, jaṭā and, for two of them, a cross-legged seated position) are said in accompanying labels to be Matsyendranātha (Figure 13), Gorakṣanātha (Figure 14) and Cauraṅginātha (Figure 15).106
Bhatt dates these images to the 14th or 15th century,107 which seems plausible and renders the ascription of a date of the 10th-century to the image of Matsyendra in the Mangalore Museum even more unlikely, since that statue has the siṅgī or horn otherwise found only in north Indian depictions of Nāth yogis from the 15th century onwards, and which is not found on the three images in the Mañjunātha temple. The statues indicate that there was previously a stronger Nāth presence at the temple than there is now.
The Āḷupa king Kundavarman, who the 1068 ce inscription says established the image of Lokeśvara, is described in the sixth verse of the inscription as pādāravindabhramaro bālacandraśikhāmaṇeḥ, “a bee at the lotus feet of he whose crest-jewel is a young moon”, i.e., a devotee of Śiva. There are no indications that the region’s primary religion was ever Buddhism; when Kadri was Buddhist it would have been an enclave within a Śaiva kingdom. Today the central shrine of the temple at Kadri contains a self-born liṅga worshipped as a form of Śiva called Mañjunātha. This name of Śiva is attested nowhere other than the Tuḷu region (and possibly its environs),108 where it is understood to mean “the gentle Lord”, mañju meaning “gentle” in Sanskrit. It is of course also redolent of Mañjuśrī (or Mañjuvajra, the form in which Mañjuśrī is found at Kadri), for which it is an alternative name,109 and it is likely that the primary focus of worship at Kadri when it was Buddhist was the Bodhisattva Mañjuvajra, a sādhana of whom was written there by Jñānapāda in the ninth century110 and an icon of whom was installed there in the eleventh century. Within two centuries of the establishment of the Lokeśvara image at the Kadirikā vihāra, the temple may already have become a Śaiva shrine: in an inscription of the Āḷupa queen Ballamahādevī who ruled 1277–92, she calls herself a worshipper of Machinātha.111 Bhatt sees this as possibly the first reference to Śiva as Mañjunātha.112 Machinātha could alternatively refer to Matsyendra, whose name is a compound of matsya, “fish”, and indra, “lord”: ma[c]chi is a middle Indic form of matsya, and nātha a synonym of indra. Two late 14th-century inscriptions at Kadri mention the worship of Mañjinātha113 and della Valle, who visited the site in 1624, says that the temple deity (which he was not allowed to see) was called “Moginato”.114 In contrast, two copper plate Tuḷu inscriptions dated 1329 ce kept inside the Kadri monastery mention Kadire Mañjunātheśvara and Kadre Śrī Mañjunāth.115
The specific referents of the name Mañjunātha and its variants as found in historical sources pertaining to Kadri are thus often uncertain, but they are likely to have referred, in sequence, to the Buddhist deity Mañjuvajra, the Nāth Matsyendra, and a localised form of Śiva.
Textual, material and epigraphic sources point to Kadri’s importance as a centre of Vajrayāna Buddhism from the early 9th to late 11th centuries, and indicate that it had become Śaiva by the early 14th century at the latest. The similarities between the statues of Lokeśvara and Matsyendra from the site indicate a direct continuity between Vajrayāna and specifically Nāth Śaiva traditions, but the Nāth presence is not confirmed by epigraphic evidence until the late 15th century, and the details of the transition from Buddhism to Śaivism are not inferrable from material and epigraphic sources. If we return to our textual sources, however, some further information may be obtained.
The c. 1400 Telugu Navanāthacaritramu of Gauraṇa tells the stories of seven of the nine Nāths of its title, including Matsyendra, who is the guru of the other eight, and Mañjunātha, the only one of the eight to be identified as the son of Matsyendra. Mañjunātha’s story takes place in Mangalore116 and is one of the earliest tellings of a famous Nāth legend found in many different sources and versions.117 Matsyendra travels to Mangalore with a group of his disciples. On their arrival the king of the city dies. His minister Prabuddha conceals his death, but Matsyendra, through his yogic sight, knows what has happened. He decides to leave his body and enter that of the king in order to experience worldly pleasures and thereby confirm that the way of the yogi is superior to worldly life. He uses a yogic technique to enter and reanimate the body of the king. Prabuddha realises what has happened but decides not to tell the people and advises Matsyendra on how to rule. Matsyendra enjoys sexual relations with his queens, and fathers a son by the chief queen. His disciples, who have been guarding his body in a mountain cave in the meantime, realise that he has forgotten himself and Gorakṣa is sent to rescue him. Gorakṣa manages to convince Matsyendra to shed all his worldly attachments with the exception of his love for his son. Gorakṣa kills the boy and the shock brings Matsyendra to his senses. He returns to his body, Gorakṣa revives the prince, and Prabuddha joins Matsyendra after arranging the king’s succession.
Matsyendra’s son is initiated as a siddha, placed on a kūrmāsana and given the name Mañjunātha. Leaving him there, Matsyendra and the other siddhas go to a cave in the Narendra mountains,118 where Prabuddha is initiated, thus becoming one of the nine Nāths of the text’s title, and given the name Buddhasiddha. Matsyendra instructs his disciples in yoga and they go on a tour of the the holy sites of north India to disseminate his teachings.
The story is in part an origin myth for the shrine of Mañjunātha at Kadri. Mañjunātha is placed on a kūrmāsana, which Śaivasiddhānta Pratiṣṭhātantras identify as the altar on which an image of a deity is to be installed.119 He is left in Kadri, while Matsyendra and his disciples continue to have adventures across India. Gorakṣa’s rescue of his guru Matsyendra from a life of debauchery is an allegory of his reformation of Nāth religious practice, i.e., the triumph of his celibate haṭhayoga over Matsyendra’s Kaula sexual rites.120 Sexual ritual was central to the practices of the Vajrayāna traditions which flourished at Kadri; the Kadri Matsyendra legend as told in the Navanāthacaritramu may also reflect the specific takeover by celibate Nāth yogis of the Kadri Vajrayāna tradition. There are faint traces of Buddhism in the Navanāthacaritramu. Gauraṇa mentions two types of yoga, a royal (rāculī) method as practised by Matsyendra, and a “peerless” (anuturya) method which Matsyendra teaches to Cauraṅgi.121 The name anuturya may be a contraction, metri causa,122 of anuttarīya, which is unattested as a name for a type of yoga in Sanskrit but which, as anuttara, has parallels in Tibetan works.123 The element buddha in the pre- and post-initiation names of the minister of the Kadri king, Prabuddha and Buddhasiddha, is also suggestive of a connection with Buddhism.
If we turn now to doctrinal texts, we find further details of the transition from Vajrayāna Buddhism to Nāth Śaivism at Kadri. This is demonstrated most clearly by the parallels between the Buddhist Amṛtasiddhi and the Śaiva Amaraughaprabodha. The Amaraughaprabodha borrows five verses from the Amṛtasiddhi and paraphrases it extensively elsewhere.124 As shown above, the Amṛtasiddhi’s place of composition is uncertain but is likely to have been in the Deccan. There is similarly no firm evidence for the place of composition of the Amaraughaprabodha, but its dependence upon the Amṛtasiddhi suggests a southern origin, which is supported by all its extant manuscripts being written in southern scripts and further corroborated by parallels with other southern texts and traditions, which will be explored below.
In addition, the new formulation of yoga taught in the Amaraughaprabodha may be traced to both the Amṛtasiddhi and tantric Buddhist traditions of haṭhayoga which are known to have been practised at Kadri. The Buddhist scholar Jñānapāda, who, as we have seen, is likely to have resided at Kadri during the first half of the ninth century, wrote a sādhana of Mañjuvajra, who is the chief deity of the Guhyasamāja in his teaching125 and is depicted in the most spectacular of the three Kadri bronzes. The Guhyasamājatantra advocates haṭhayoga, albeit as a fallback method of achieving awakening.126 The Amaraughaprabodha draws on this understanding of the term haṭhayoga when it uses it as the name of the yoga method of the Amṛtasiddhi, which, it states, is subordinate to rājayoga. As suggested by the primacy of celibate practice implied by the Navanāthacaritramu’s legend of Matsyendra’s rescue by Gorakṣa (who, like all his co-disciples, is avowedly celibate), the Amaraughaprabodha teaches that the control of bindu, i.e., semen, is central to haṭhayoga. It refers to the practice of vajrolimudrā, a method of controlling semen which is likely to have been part of certain traditions of Vajrayāna sexual ritual, but makes no mention of its use in such rites, saying that in the Amara tradition it is equanimity and the flowing of the breath in the central channel.127 This transition from Vajrayāna to Śaiva haṭha practice evident in the Amaraughaprabodha is echoed in the story from the Līḷācaritra referred to above,128 in which Virūpa, the first teacher of the yoga of the Amṛtasiddhi, is said to switch allegiance from the Vajra to Amara lineage after his conversion by Gorakṣa. The polyvalence of the term ogha in the name of the Amaraughaprabodha (whose teachings are attributed to Gorakṣa), allows it to mean both “Awakening (prabodha) by the Stream (ogha) of the Nectar of Immortality (amara)” and “Awakening in the Amara lineage (ogha)”.
The strongest indication of a link between Kadri and the Amaraughaprabodha is found in the siddhas it invokes. It opens with a maṅgala verse to Ādinātha, Mīnanātha,129 Cauraṅginātha and Siddhabuddha, and in the next verse ascribes its teachings to Gorakṣa.130 Ādinātha is the name of the form of Śiva to which many Nāth lineages trace their origin. The triad of Mīna, Gorakṣa and Cauraṅgi has primacy only in Nāth lineages from the Deccan and Konkan. Cauraṅgi rarely features in Nāth traditions from further north131 and no Sanskrit haṭha texts apart from the Amaraughaprabodha, Haṭhapradīpikā and derivative works mention him. Matsyendra, Gorakṣa and Cauraṅgi are the main protagonists in both the Navanāthacaritramu and Kadalīmañjunāthamāhātmya.132 The invocation of these three Nāths thus supports a southern origin for the Amaraughaprabodha, with Kadri a strong candidate not only because they are central to its māhātmya, but also because, as we have seen, the same triad adorns the main shrine of Mañjunātha.
The mention of Siddhabuddha in the Amaraughaprabodha’s maṅgala verse points more specifically to Kadri as the location of that text’s composition. Unlike the other three siddhas named in the verse, Siddhabuddha is little known; within the Sanskrit haṭha corpus his name is otherwise found only in the list of 29 siddhas in the Haṭhapradīpikā and texts deriving from it.133 The Haṭhapradīpikā borrows twenty verses directly from the Amaraughaprabodha and may have taken Siddhabuddha’s name from its maṅgala verse. Arguing against this possibility, however, is that in the Haṭhapradīpikā’s list of siddhas, Siddhabuddha is not grouped with the other three siddhas named in the Amaraughaprabodha’s maṅgala verse. A more likely source, or milieu, for the Haṭhapradīpikā to have taken the name Siddhabuddha from is suggested by the only other texts known to me in which he is mentioned. These are two Telugu works, the c. 1400 ce Navanāthacaritramu and the contemporaneous Vikramārkacaritramu of Jakkana. The Haṭhapradīpikā has several parallels with the Navanāthacaritramu, in particular in the names of the siddhas found in both texts,134 some of which are otherwise obscure. As we have seen, the Navanāthacaritramu includes Buddhasiddha among the nine Nāths of its title. That this Buddhasiddha is simply Siddhabuddha with the elements of his name transposed is indicated by his name being found as Siddhabuddha in the Vikramārkacaritramu, which gives a list of nine Nāths almost identical to that of the Navanāthacaritramu (Mañjunātha is replaced by Ādinātha). A further detail supports the identification of the Siddhabuddha of the Amaraughaprabodha with the Buddhasiddha/Siddhabuddha of the Navanāthacaritramu and Vikramārkacaritramu. In the Navanāthacaritramu, before his initiation Buddhasiddha is Prabuddha, the crafty minister of the king of Mangalore. The Vikramārkacaritramu highlights a quality of each of the nine Nāths. Siddhabuddha comes after Ādinātha, Matsyendra, Sāraṅga (the pre-initiatory name of Cauraṅgi) and Gorakṣa, and is praised for his intellect.135 In the Amaraughaprabodha’s maṅgala verse Siddhabuddha is the only siddha to have an epithet, which is dhīmate, “the wise one”. These parallels allow for a tentative identification of the Siddhabuddha of the Amaraughaprabodha’s maṅgala verse with the Buddhasiddha of the Navanāthacaritramu, who resided at Kadri.

5. Final Remarks

Our available evidence indicates that the transition from Buddhism to Śaivism at Kadri was peaceful. No textual sources related to Kadri suggest a violent takeover and the Vajrayāna bronzes at the site remained more or less in situ without being damaged.136 Furthermore, the later Nāth statues of Matsyendra are modelled on that of the Buddhist Lokeśvara, indicating direct continuity between the Nāth tradition at Kadri and its Vajrayāna forerunner. Other than the icon on Matsyendra’s crown of matted locks, however, all material traces of Buddhism were removed, or, as in the case of the three bronzes, reassigned as Hindu. This is true also for the Amaraughaprabodha, which, in adopting and adapting the teachings of the Amṛtasiddhi, removes features that are obviously Buddhist but leaves traces of more obscure Buddhist doctrines which were not recognised by its redactors.137
The Śaiva tradition which took over at Kadri thus absorbed elements of the earlier Vajrayāna tradition but tried to efface anything overtly Buddhist. This is in contrast with the Newar tradition in the Kathmandu Valley, where Avalokiteśvara/Lokeśvara and Matsyendra have been identified since at least the 15th century, and Buddhism and Śaivism remain inextricably entangled.138 19th-century chronicles claim an 11th-century Karnatakan origin for the Newar kings and mention connections with the Konkan,139 hinting at a possible origin in south India for the Newari identification of Avalokiteśvara and Matsyendra, and at links with the Kadri tradition, but the earliest record of their identification in Nepal whose date is certain is from the 15th century.140 Buddhism’s survival in the Kathmandu Valley is in stark contrast with its almost complete demise in India. There is no clear reason for either of these outcomes, but the Kathmandu Valley’s isolation was responsible for other divergences from the history of the subcontinent, such as its not being subjugated by the Mughals or British, and is likely also to be at least partly responsible for Buddhism’s survival there.
The apparently peaceful transition at Kadri does not support arguments that Brahmanism or Islam violently removed Buddhism from India.141 Textual and epigraphical sources do however suggest that interactions between tantric Buddhists and Śaivas in the region of Kadri were not always friendly. Desai notes two late 12th-century south Indian inscriptions from Karnataka which indicate that Śaiva aggression may have been responsible for the destruction of Buddhist vihāras.142 And, as we have seen above, several texts tell the story of Virūpa himself destroying a śivaliṅga at Drākṣārāma.143 The Nityāhnikatilaka, a text of the Paścimāmnāya Kaula tradition from which some Nāth lineages developed, reports that Unmanīśanātha destroyed 125,000 Buddhist vihāras in heaven, as a result of which nine Buddhists fell to earth and were subsequently initiated by him.144
But these hints of a violent transfer are exceptional. Other sources point to debate or instruction as the means by which one doctrine superseded the other. Unsurprisingly, the Śaiva sources at our disposal almost always tell of Buddhists capitulating to Śaiva arguments. Of particular relevance to the topic in hand is a story in the Līḷācaritra of Virūpa, after his conversion by Gorakṣa from Buddhism to Śaivism,145 defeating in debate kṣapaṇakas, i.e., Jain or Buddhist ascetics, and re-establishing the Vedic religion at Ellora.146 Kularatnoddyota 11.31 includes the Buddha (and the Jina and Matsyendra as well as many others) in a list of siddhas who were taught the Kaulajñāna of the Pūrvāmnāya by Mitranātha.147 I know of one instance in which the transmission went the other way, suggesting a willingness by Śaiva traditions to accept at least some Buddhist teachings: the Ṣaṭsāhasrasaṃhitā, a Paścimāmnāya Kaula Śaiva text, tells the story of eight Nāths being given teachings by a teacher whose name, Vajrabodhi, indicates a Buddhist affiliation.148
The demise of Buddhism, at Kadri and elsewhere, thus seems to have usually been peaceful. It may in fact have been imperceptible to all but a few. Boundaries between Śaiva and Vajrayāna siddha traditions were porous. Some siddhas were clearly originally either Śaiva or Vajrayāna,149 but others cannot be tied to one or other tradition.150 The Amṛtasiddhi has some features which definitively identify it as Buddhist, such as a metaphysics that has only four basic elements, not the five of Hindu traditions, but it is also peculiarly Śaiva for a Buddhist work, including such un-Buddhist notions as jīvanmukti, liberation while living.151 Patronage and devotion were similarly non-partisan. We have seen how the Kadri Lokeśvara inscription records the installation of a Buddhist icon by a Śaiva king. Further north in the subcontinent, some Gāhaḍavāla and Pāla monarchs had both Buddhist and Kaula gurus.152 Thus there was a shared siddha world, especially of praxis, from which, in India, the Śaiva siddha tradition emerged, mostly free from Buddhist vestiges.
The above evidence for the existence of Vajrayāna Buddhism in the Konkan is compelling, but scant. Vajrayāna was never a state religion in the Tuḷu region, whose rulers were Śaiva, but the 1068 ce Kadri inscription shows that it was patronised by at least one Tuḷu monarch. It may also have had patrons from further afield. Gomin Avighnakāra, a Buddhist devotee from Gauḍa, made an endowment at Kanheri in the northern Konkan in 854 ce153 and it is not impossible that Kadri was also supported by patrons from Gauḍa. But there is much more evidence for a Tamil connection. Unlike all other inscriptions at Kadri, which are written in the Kannada script and make no reference to Buddhism, the 1068 ce inscription recording the installation of Lokeśvara is in Grantha, the script used in the Tamil region for writing Sanskrit. In the c. 13th-century Matsyendrasaṃhitā, which was composed in either the Tamil or Konkan regions,154 Gorakṣa is a Coḷā king before he becomes a yogi.155 The Tamil Tirumantiram, whose composition, like the installation of Lokeśvara at Kadri, dates to the second half of the 11th century ce,156 has both Buddhist and Śaiva elements in its yoga, and several features of its cult are shared with the Kadri yogi tradition.157 The bronzes of Lokeśvara and Mañjuvajra at Kadri are similar to contemporaneous Coḷā pieces.158 The Kadri Buddha may also be from the Tamil region, or perhaps Sri Lanka or southeast Asia, bringing to mind Mangalore’s importance as a trading port (whose cosmopolitanism in the 12th-century has been vividly reimagined by Amitav Ghosh).159 A connection between Kadri and Sri Lanka is supported by the inclusion of Matsyendranātha among a list of eight forms of Avalokiteśvara (who in Sri Lanka is also commonly known as Nātha) described in the Śariputra, a text of uncertain date used by the makers of images in Sri Lankan temples.160 In addition, the Guhyasamāja exegete Jayabhadra, who spent some time in the Konkan before travelling to Vikramaśīla, was originally from Sri Lanka.161 Kadri was at the western extremity of a Buddhist maritime world whose hub was Sri Lanka and which extended to southeast Asia. The possibility of Nāth influence on the mixed Śaiva-Buddhist traditions of Java and Bali is hinted at by the presence of an image of Matsyendra on a statue of Cāmuṇḍī in Java dated 1292.162 The links between Kadri and the Tamil sphere of influence point to Kadri being supported by the Coḷā empire, which would have been connected to Mangalore by sea and land: the most important centres of Buddhism in the Coḷā empire were the ports of Kanchi and Nagapattinam, and its territory reached near Mangalore until its demise in the 14th century. Like Kundavarman, who established the Lokeśvara image at Kadri, certain Coḷā rulers such as Rājarāja I (pp. 985–1014 ce) were Śaivas who also patronised Buddhism.163 The Kadri vihāra may thus have benefited from Coḷā patronage as well as that of its local overlords. When these sources of support dried up, either as the result of increasingly exclusive patronage of Śaivism by both the Āḷupa164 and Coḷā kings or the demise of the Coḷā empire, the Kadri Buddhist tradition was left isolated and unsupported.
Padmanabh Jaini, analysing the very different historical outcomes for Buddhism and Jainism in India and citing the example of Kadri, argues convincingly that Hinduism was able to assimilate Buddhism — and not Jainism — because the Bodhisattvas that were central to Buddhist devotional cults could be reconfigured as Śaiva deities and siddhas.165 The Vajrayāna cult at Kadri had the Bodhisattva Mañjuvajra as its central deity, hence Jñānapāda’s writing a sādhana of him there in the ninth century and the name of the temple being Mañjunātha, an alternative for Mañjuvajra. But then the Śaiva king Kundavarman established an image of the Bodhisattva Lokeśvara in the temple. Lokeśvara’s iconography is closely related to that of Śiva — with whom he shares a dual role as ruler and ascetic yogi166 — and thus also to that of Matsyendra, who subsequently became the central object of worship at the Kadri monastery. Legend has it that Matsyendra used the yogic technique of parakāyapraveśa, entry into another’s body, to revive the dead king of Kadri; historical sources indicate that Kundavarman’s act of inclusive benevolence at Kadri paved the way for a moribund tantric Buddhism to use the same technique to achieve the yogic aim of kālavañcana, cheating death, by entering both the iconic body of Matsyendra and the corpus of teachings attributed to his putative disciple Gorakṣa.


The research for this article was carried out as part of the Hatha Yoga Project ( This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 647963).


I would like to thank Véronique Bouillier, whose pioneering studies on Kadri first brought it to scholarly attention, for her support and encouragement. I would also like to thank the following for their help with this article: Andrea Acri, Amol Bankar, Daniela Bevilacqua, Jason Birch, Crispin Branfoot, Lucy May Constantini, Manu Devadevan, Rich Freeman, Dominic Goodall, Viswanatha Gupta, John Guy, Jacqueline Hargreaves, Shaman Hatley, Ken Ishikawa, Shreekant Jadhav, the late Ānandanāth Jogī, Jamal Jones, Christian Luczanits, Lubomír Ondračka, Gethin Rees and Péter-Dániel Szántó.
Religions 10 00273 i001


The following abbreviations are used in this manuscript:
nakNational Archives Kathmandu
ngmppNepal-German Manuscript Preservation Project


  1. Primary Sources

    Amaraughaprabodha, ongoing critical edition by Jason Birch.
    Amṛtasiddhi, ongoing critical edition by James Mallinson and Péter-Dániel Szántó.
    Kadalīmañjunāthamāhātmyam, ed. Śambhu Śarmā Kaḍava. Kāśī: Gorakṣa Ṭilla Yoga Pracāriṇī. 1957.
    Gopālarājavaṃśāvalī, ed. and tr. by Dhanavajra Vajrācārya and Kamal P. Malla. Nepal Research Centre Publication 9. Stuttgart: Steiner. 1985.
    Grub thob brgyad bcu rtsa bzhi’i lo rgyus, see Robinson 1979.
    Tārārahasyam of Brahmānanda, ed. Jīvānandavidyāsāgara Bhaṭṭācārya. Calcutta: Calcutta Press. 1896.
    Navanāthacharitra of Gauraṇa, ed. K. Ramakrishnaiya. Madras University Telegu Series No. 7. Madras. 1937.
    Nityāhnikatilaka. nak 3–384, ngmpp A41/11; palm-leaf; dated c. 1452.
    Nepālabhūpavaṃśāvalī. See Bajracharya and Michaels 2015.
    Bodhisattvabhūmi, ed. Wogihara. 1930–1936. Tokyo: Sankibo.
    Yogajāgama Institut français de Pondichéry transcript T0024.
    Virūpākṣapañcāśikā with the commentary of Vidyācakravartin, ed. T. Gaṇapati Śāstrī. Trivandrum Sanskrit Series 9. Trivandrum. 1910.
    Vivekadarpaṇa, ed. V.D. Kulkarni. Marathi Journal Vol. 6, pp. 89-129. Osmania University: Hyderabad. 1971.
    Ṣaṭsāhasrasaṃhitā. ngmpp b32–6.
    Sahasrāgama Institut français de Pondichéry transcript T0033.
    Suprabhedāgama. Published 1931 ce by Devīkoṭaparipālanasaṃgha, editor not named. Transcription available at
    Suvarṇaprabhāsottama, ed. Prods Oktor Skjærvø in This most excellent shine of gold, king of kings of sūtras: The Khotanese Suvarṇaprabhāsottama. 2 Vols. Harvard University. 2004.
    Sekoddeśa, ed. Giacomella Orofino. Rome: Istituto italiano per il Medio ed Estremo Oriente. 1994.
    South Indian Inscriptions (Texts) Volume VII. Miscellaneous Inscriptions from the Tamil, Malayalam, Telugu and Kannada Countries, ed. K.V. Subrahmanya Aiyer. Calcutta: Government of India. 1932.
    Haṭhapradīpikā of Svātmārāma, ed. Svāmī Digambarjī and Dr Pītambar Jhā. Lonavla: Kaivalyadhām S.M.Y.M. Samiti. 1970.
    Hevajratantra, ed. David L. Snellgrove in The Hevajra Tantra, A Critical Study. Part II Sanskrit and Tibetan Texts. London: OUP. 1959.
  2. Secondary Sources

  3. Acri, Andrea. 2015. Revisiting the Cult of “Śiva-Buddha” in Java and Bali. In Buddhist Dynamics in Premodern and Early Modern Southeast Asia. Singapore: Institute of Southeast Asian Studies, pp. 261–82. [Google Scholar]
  4. Acri, Andrea. 2016. Esoteric Buddhism in Mediaeval Maritime Asia. Networks of Masters, Texts, Icons. Singapore: ISEAS—Yusof Ishak Institute. [Google Scholar]
  5. Acri, Andrea. 2018. Maritime Buddhism. Oxford Research Encyclopedia of Religion. Available online: (accessed on 14 April 2019).
  6. Badger, George Percy. 1863. The Travels of Ludovico di Varthema in Egypt, Syria, Arabia Deserta and Arabia Felix, in Persia, India and Ethiopia, A.D. 1503 to 1508. Translated from the original Italian Edition of 1510, with a Preface, by John Winter Jones, Esq., F.S.A., and edited, with notes and an introduction, by George Percy Badger. London: The Hakluyt Society. [Google Scholar]
  7. Baker, Ian. 2018. Tibetan Yoga: Somatic Practice in Vajrayāna Buddhism and Dzogchen. In Yoga in Transformation. Vienna: V& R Unipress, pp. 405–60. [Google Scholar]
  8. Bajracharya, Manik, and Axel Michaels. 2016. Nepālikabhūpavaṃśāvalī. History of Kings of Nepal. A Buddhist Chronicle. Introduction and Translation. Kathmandu: Himal Books. [Google Scholar]
  9. Bankar, Amol. 2013. Two Inscriptions from the Ghorāvadeśvara Caves near Shelarwadi. In Annals of the Bhandarkar Oriental Research Institute 2011 (Volume XCII). Pune: Bhandarkar Oriental Research Institute, pp. 93–102. [Google Scholar]
  10. PG Bhatt. 1975. Studies in Tuḷuva History and Culture. Manipal: P.G. Bhatt. [Google Scholar]
  11. Birch, Jason. 2011. The Meaning of haṭha in early Haṭhayoga. Journal of the American Oriental Society 131: 527–54. [Google Scholar]
  12. Birch, Jason. Forthcoming. The Amaraughaprabodha: New Evidence on its Manuscript Transmission. In Śaivism and the Tantric Traditions. Leiden: Brill.
  13. Bisschop, Peter. 2018. Buddhist and Śaiva Interactions in the Kali Age: The Śivadharmaśāstra as a Source of the Kāraṇḍavyūhasūtra. Indian-Iranian Journal 61: 396–410. [Google Scholar] [CrossRef]
  14. Bopearachchi, Osmund. 2014. Sri Lanka and Maritime Trade: Bodhisattva Avalokiteśvara as the Protector of Mariners. In Asian Encounters: Exploring Connected Histories. London: OUP, pp. 162–87. [Google Scholar]
  15. Bouillier, Véronique. 2008. Itinérance et vie Monastique. Les Ascètes Nāth Yogīs en Inde Contemporaine. Paris: Éditions de la Maison des Sciences de L’homme. [Google Scholar]
  16. Bouillier, Véronique. 2009. The Pilgrimage to Kadri Monastery (Mangalore, Karnataka): A Nāth Yogī Peformance. In Patronage and Popularisation, Pilgrimage and Procession. Channels of Transcultural Translation and Transmission in Early Modern South Asia. Papers in Honour of Monika Horstmann. Studies in Oriental Religions 58. Wiesbaden: Harrassowitz Verlag, pp. 135–46. [Google Scholar]
  17. Briggs, George Weston. 1938. Gorakhnāth and the Kānphaṭa Yogīs. Delhi: Motilal Banarsidass. [Google Scholar]
  18. Chattopadhyaya, Debiprasad. 1970. Tāranātha’s History of Buddhism in India. Simla: Indian Institute of Advanced Study. [Google Scholar]
  19. Cowell, Edward Byles, and Julius Eggeling. 1876. Catalogue of Buddhist Sanskrit Manuscripts in the Possession of the Royal Asiatic Society (Hodgson Collection). The Journal of the Royal Asiatic Society of Great Britain and Ireland 8: 1–52. [Google Scholar] [CrossRef]
  20. Dasgupta, Shashibhushan. 1946. Obscure Religious Cults, 3rd ed. Calcutta: Firma KLM Private Ltd. [Google Scholar]
  21. Davidson, Ronald. 2002. Indian Esoteric Buddhism: A Social History of the Tantric Movement. New York: Columbia University Press. [Google Scholar]
  22. Davidson, Ronald M. 2005. Tibetan Renaissance: Tantric Buddhism in the Rebirth of Tibetan Culture. New York: Columbia University Press. [Google Scholar]
  23. Desai, Prasannavadan Bhagwanji. 1954. Buddhist Antiquities in Karnataka. Journal of Indian History 32: 85–92. [Google Scholar]
  24. Deshpande, Madhusudan Narhar. 1986. The Caves of Panhāle-Kāji (Ancient Pranālaka); New Delhi: Archaeological Survey of India.
  25. Dowman, Keith. 1985. Masters of Mahamudra. Songs and Histories of the Eighty-Four Buddhist Siddhas. New York: State University of New York Press. [Google Scholar]
  26. Freeman, John R. 2006. Shifting Forms of the Wandering Yogi: The Teyyam of Bhairavan. In Masked Ritual and Performance in South India: Dance, Healing and Possession. Ann Arbor: The University of Michigan, pp. 147–83. [Google Scholar]
  27. Ghosh, Amitav. 1994. In an Antique Land. London: Granta. [Google Scholar]
  28. Gokhale, Shobhana. 1990. Kanheri Inscriptions. Pune: Deccan College Post Graduate and Research Institute. [Google Scholar]
  29. Goodall, Dominic. 1998. Bhaṭṭa Rāmakaṇṭha’s Commentary on the Kiraṇatantra. Volume 1: Chapters 1–6. Critical Edition and Annotated Translation. Pondicherry: Publications de l’Institut français d’Indologie. [Google Scholar]
  30. Goodall, Dominic. 2004. The Parākhyatantra. A Scripture of the Śaiva Siddhānta. A Critical Edition and Annotated Translation. Pondicherry: Publications de l’Institut français d’Indologie. [Google Scholar]
  31. Goodall, Dominic. 2011. The Throne of Worship: An ‘Archaeological Tell’ of Religious Rivalries. Studies in History 27: 221–50. [Google Scholar] [CrossRef]
  32. Grey, Edward. 1892. The Travels of Pietro della Valle in India. From the old English Translation of 1664 by G. Havers. London: The Hakluyt Society. [Google Scholar]
  33. Grunwedel, Albert. 1916. Die Geschichten der vierundachtzig Zauberer (Mahāsiddhas) aus dem Tibetischen übersetzt. Leipzig: Teubner. [Google Scholar]
  34. Hatley, Shaman. 2016. Converting the Dākinī: Goddess Cults and Tantras of the Yoginīs between Buddhism and Śaivism. In Tantric Traditions in Transmission and Translation. New York City: OUP, pp. 37–86. [Google Scholar]
  35. Hatley, Shaman. 2018. The Brahmayāmalatantra or Picumata, Volume I, Chapters 1–2, 39–40 and 83. Revelation, Ritual, and Material Culture in an Early Śaiva Tantra. Pondicherry: École française d’Extreme-Orient. [Google Scholar]
  36. Isaacson, Harunaga. 2008. Himalayan Encounter: the teaching Lineage of the Marmopadeśa. Studies in the Vanaratna Codex 1: 2–6. [Google Scholar]
  37. Isaacson, Harunaga, and Sferra Francesco. 2014. The Sekanirdeśa of Maitreyanātha (Advayavajra) with the Sekanirdeśapañjikā of Rāmapāla. Critical Edition of the Sanskrit and Tibetan Texts with English Translation and Reproductions of the MSS. Serie Orientale Roma fondata da Giuseppe Tucci Vol. CVII. Napoli: Universita degli Studi di Napoli “L’Orientale”. [Google Scholar]
  38. Jaini, Padmanabh. 1980. The Disappearance of Buddhism and the Survival of Jainism: A Study in Contrast. In Studies in History of Buddhism. Delhi: B.R. Publishing Corporation, pp. 81–91. [Google Scholar]
  39. Jones, Jamal André. 2018. A Poetics of Power in Andhra, 1323–1450 ce. Doctoral dissertation, Faculty of the Division of the Humanities, University of Chicago, Chicago, IL, USA. [Google Scholar]
  40. Kandahjaya, Hudaya. 2016. Saṅ Hyaṅ Kamahāyānikan, Borobudur, and the Origins of Esoteric Buddhism in Indonesia. In Esoteric Buddhism in Mediaeval Maritime Asia. Networks of Masters, Texts, Icons. Singapore: ISEAS—Yusof Ishak Institute, pp. 67–112. [Google Scholar]
  41. Kasthuri, K. Anusha. 2016. Preliminary Investigation of Sri Lankan Copper-alloy Statues. STAR: Science & Technology of Archaeological Research 2: 159–76. [Google Scholar]
  42. Kim, Jinah. 2014. Local Visions, Transcendental Practices: Iconographic Innovations of Indian Esoteric Buddhism. History of Religions 54: 34–68. [Google Scholar] [CrossRef]
  43. Kiss, Csaba. Forthcoming. Matsyendrasaṃhitā. Pondicherry: École française d’Extreme-Orient.
  44. Kvaerne, Per. 1977. An Anthology of Buddhist Tantric Songs: A Study of the Caryāgīti. Oslo: Universitetsforlaget. [Google Scholar]
  45. Lévi, Sylvain. 1905. Le Népal: Étude Historique d’un Royaume Hindou. Paris: Ernest Leroux, vol. 1. [Google Scholar]
  46. Locke, John K. 1980. Karunamaya. The Cult of Avalokitesvara-Matsyendranath in the Valley of Nepal. Kathmandu: Sahayogi Prakashan. [Google Scholar]
  47. Luczanits, Christian. 2006. The Eight Great Siddhas in Early Tibetan Painting. In Holy Madness: Portraits of Tantric Siddhas. New York: Rubin Museum of Art and Chicago: Serindia Publications, pp. 76–91. [Google Scholar]
  48. Malalasekera, Gunapala Piyasena. 1937. Dictionary of Pali Proper Names. London: Luzac & Co. [Google Scholar]
  49. Malla, Shiva Raj Shreshtha. 2006. Nanyadeva, His Ancestors and Their Abhijana (Original Homeland). Ancient Nepal 160: 1–20. [Google Scholar]
  50. Mallinson, James. 2011. Nāth Saṃpradāya. In Brill Encyclopedia of Hinduism. Leiden: Brill, vol. 3, pp. 407–28. [Google Scholar]
  51. Mallinson, James. 2013. Yogic Identities: Tradition and Transformation. Smithsonian Institute Research Online. Available online: (accessed on 14 April 2019).
  52. Mallinson, James. 2014. Hathayoga’s Philosophy: A Fortuitous Union of Non-Dualities. Journal of Indian Philosophy 42: 225–47. [Google Scholar] [CrossRef]
  53. Mallinson, James. Forthcoming a. The Amṛtasiddhi: Haṭhayoga’s tantric Buddhist source text. In Śaivism and the Tantric Traditions. Leiden: Brill.
  54. Mallinson, James. Forthcoming b. Yoga and Yogis: the Texts, Techniques and Practitioners of Early Haṭhayoga. Pondicherry: EFEO.
  55. Mathew, Kalloor Mathen. 1988. History of the Portuguese Navigation in India, 1497–1600. Delhi: Mittal Publications. [Google Scholar]
  56. Michaels, Axel. 1985. On 12th-13th Century relations between Nepal and South India. Journal of the Nepal Research Centre VII: 69–73. [Google Scholar]
  57. Muñoz, Adrián. 2011. Matsyendra’s ‘Golden Legend’: Yogi Tales and Nath Ideology. In Yogi Heroes and Poets: Histories and Legends of the Nāths. New York: State University of New York Press, pp. 109–28. [Google Scholar]
  58. Nagaraju, S. 1969. A Rare Śaivo-Buddhist Work with Vaishnavite Interpolations. In Kannada Studies. Mysore: Institute of Kannada Studies, vol. 3, pp. 67–75. [Google Scholar]
  59. Nagaraju, S. 1983. Vestiges of Buddhist Art. Marg 35: 5–14. [Google Scholar]
  60. Newman, John. 1998. The Epoch of the Kālacakratantra. Indo-Iranian Journal 41: 319–49. [Google Scholar] [CrossRef]
  61. Ondračka, Lubomír. 2011. What should Mīnanāth do to save his life? In Yogi Heroes and Poets: Histories and Legends of the Nāths. New York: State University of New York Press, pp. 129–42. [Google Scholar]
  62. Pai, M. Govind. 1946. The Date of Gorakhanātha. New Indian Antiquary 8: 57–66. [Google Scholar]
  63. Paranavitana, Senarat. 1928. Mahāyānism in Ceylon. Ceylon Journal of Science Section G.—Archaeology, Ethnology, Etc. 2: 35–71. [Google Scholar]
  64. Pathak, K. B. 1912. The Ajivikas, a Sect of Buddhist Bhikshus. The Indian Antiquary 41: 88–90. [Google Scholar]
  65. Raeside, Ian M. P. 1982. Dattātreya. Bulletin of the School of Oriental and African Studies 45: 489–500. [Google Scholar] [CrossRef]
  66. Rawat, Yadubir Singh. 2009. Hill Fort of Anarta: Discovery of a Unique Early Historical Fort with Cave Dwellings, Buddhist Idols and Remains at Taranga in North Gujarat. Puratattva 39: 96–106. [Google Scholar]
  67. Rees, Gethin. Forthcoming. Rock-cut Buddhist Monasteries of the Southern Konkan. In Pratnamaṇi Felicitation Volume of Prof. Vasant Shinde. New Delhi: B.R. Publishing House.
  68. Reinelt, Kurt Joachim. 2000. Das Vivekadarpaṇa. Textanalyse und Erlauterungen zur Philosophie und praktischen Erlösungslehre der Nāthayogīs in Mahārāṣṭra. Heidelberg: Inauguraldissertation zur Erlangung des Doktorgrades der Fakultät fïr Orientalistik und Altertumswissenschaft der Universität Heidelberg. [Google Scholar]
  69. Robinson, James B. 1979. Buddha’s Lions: The Lives of the Eighty-Four Siddhas. Berkeley: Dharma Publishing. [Google Scholar]
  70. Saletore, Bhasker Anand. 1936. Ancient Karnāṭaka Vol.1 History of Tuḷuva. Poona Oriental Series No.53. Poona: Oriental Book Agency. [Google Scholar]
  71. Saletore, Bhasker Anand. 1937. The Kānaphāṭa [sic] Jogis in Southern History. The Poona Orientalist 1: 16–22. [Google Scholar]
  72. Sanderson, Alexis. 1994. Vajrayāna: Origin and Function. In Buddhism into the Year 2000. International Conference Proceedings. Bangkok/Los-Angeles: Dhammakaya Foundation, pp. 87–102. [Google Scholar]
  73. Sanderson, Alexis. 2001. History through Textual Criticism in the study of Śaivism, the Pañcarātra and the Buddhist Yoginītantras. In Les Sources et le temps. Sources and Time: A Colloquium, Pondicherry, 11–13 January 1997. Edited by François Grimal. Publications du DéPartement d’Indologie 91. Pondicherry: Institut Français de Pondichéry/École Française d’Extrême-Orient, pp. 1–47. [Google Scholar]
  74. Sanderson, Alexis. 2005. Religion and the State: Initiating the Monarch in Śaivism and the Buddhist Way of Mantras. Unpublished Article. [Google Scholar]
  75. Sanderson, Alexis. 2009. The Śaiva Age. In Genesis and Development of Tantrism. Tokyo: Institute of Oriental Culture, University of Tokyo, pp. 41–349. [Google Scholar]
  76. Sanderson, Alexis. 2011. Śaivism, Society and the State. Unpublished paper. [Google Scholar]
  77. Sanderson, Alexis. 2012. The Śaiva Literature. Journal of Indological Studies, 1–113, Numbers 24 and 25 (2012–2013). [Google Scholar]
  78. Sarde, Vijay. 2014. Nath-Siddhas [sic] Sculptures on the Someshwara Temple at Pimpri-Dumala in Pune. Bulletin of the Deccan College 74: 1–10. [Google Scholar]
  79. Sarde, Vijay. 2016. Art and Architecture of Someshwar Temple at Pimpri-Dumala, Pune District, Maharashtra. Historicity Research Journal 2: 1–22. [Google Scholar]
  80. Sarde, Vijay. 2017. Paścimī Dakkhan se prāpt Matsyendranāth aur Gorakṣanāth ke nav anveṣit Mūrti-Śilp. Arṇava 6: 94–113. [Google Scholar]
  81. Sastri, K. A. Nilakanta. 1939. Foreign Notices of South India: From Megasthenes to Ma Huan. Madras: University of Madras. [Google Scholar]
  82. Schaeffer, Kurtis R. 2002. The Attainment of Immortality: From Nāthas in India to Buddhists in Tibet. Journal of Indian Philosophy 30: 515–33. [Google Scholar] [CrossRef]
  83. Scheurleer, Pauline Lunsingh. 2008. The well-Kknown Javan statue in the Tropenmuseum, Amsterdam, and its place in Javanese sculpture. Artibus Asiae 68: 287–332. [Google Scholar]
  84. Schoterman, Jan Anthony. 1975. Some remarks on the Kubjikāmatatantra. ZDMG Supplement III 2: 932–40. [Google Scholar]
  85. Sen, Sukumar. 1956. The Nātha Cult. In The Cultural Heritage of India. Vol. 4: The Religions. Calcutta: The Ramakrishna Mission Institute of Culture, pp. 280–90. [Google Scholar]
  86. Seyfort Ruegg, David. 2001. A Note on the Relationship between Buddhist and ‘Hindu’ Divinities in Buddhist Literature and Iconology: The laukika/lokottara Contrast and the Notion of an Indian ‘Religious Substratum’. In Le parole e i marmi. Studi in onore di Raniero Gnoli nel suo 70 compleanno. Rome: Istituto per l’Africa e l’Oriente, pp. 735–42. [Google Scholar]
  87. Sferra, Francesco. 2003. Some Considerations on the Relationship between Hindu and Buddhist Tantras. In Buddhist Asia 1, Papers from the First Conference of Buddhist Studies Held in Naples in May 2001. Kyoto: Italian School of East Asian Studies, pp. 57–84. [Google Scholar]
  88. Shah, Umakant Premanand. 1957. Nāth Siddhoṃ kī Prācīn Śilpamūrtiyāṃ. Nāgarīpracāriṇī Patrikā varṣa 62: 174–207. [Google Scholar]
  89. Shetti, B.V. 1988. Bronzes of Karnataka and Andhra Pradesh. In The Great Tradition. Indian Bronze Masterpieces. New Delhi: Brijbasi, pp. 70–87. [Google Scholar]
  90. Sompura, Kantilal F. 1968. The Structural Temples of Gujarat. Ahmedabad: Gujarat University. [Google Scholar]
  91. Sompura, Kantilal F. 1969. Buddhist Monuments and Sculptures in Gujarat-A Historical Survey. Hoshiarpur: Vishveshvaranand Institue. [Google Scholar]
  92. Szántó, Péter-Dániel. 2012. Selected Chapters from the Catuṣpīṭhatantra. Doctoral thesis, University of Oxford, Oxford, UK. [Google Scholar]
  93. Szántó, Péter-Dániel. 2015. Early Works and Persons Related to the So-called Jñānapāda School. Journal of the International Association of Buddhist Studies 36: 537–62. [Google Scholar]
  94. Szántó, Péter-Dániel. 2016a. Some unknown/unedited fragments of Cakrasaṃvara literature. Paper presented at Conference ‘Transgression and Encounters with the Terrible in Buddhist and Śaiva Tantras’, Zurich, Switzerland, February 19–20. [Google Scholar]
  95. Szántó, Péter-Dániel. 2016b. A Brief Introduction to the “Amṛtasiddhi, ongoing critical edition by James Mallinson and Péter-Dániel Szántó”. Paper presented at Sanskrit Texts on Yoga, A Manuscript Workshop, London, UK, September 12–16. [Google Scholar]
  96. Szántó, Péter-Dániel. 2017. Tantric Buddhist Communities and Seeking Patronage at Medieval Indian Courts. Cambridge: Harvard University. [Google Scholar]
  97. Szántó, Péter-Dániel. Forthcoming. Siddhas. In Brill’s Encyclopedia of Buddhism. Leiden: Brill.
  98. Templeman, David. 1983. The Seven Instruction Lineages by Jonang Tāranātha. Dharamsala: Library of Tibetan Works and Archives. [Google Scholar]
  99. Templeman, David. 1995. The Origin of Tārā Tantra: Sgrol ma‘i Rgyud kyi Byung khung Gsal Bar Byed pa‘i lo Rgyus Gser gyi Phreng ba zhes Bya ba. Dharamsala: Library of Tibetan Works and Archives. [Google Scholar]
  100. Templeman, David. 1997. Buddhaguptanātha: A Late Indian Siddha in Tibet. In Tibetan Studies. Proceedings of the 7th Seminar of the International Association of Tibetan Studies, Graz 1995. Wien: Verlag der Österreichischen Akademie der Wissenschaften, vol. II, pp. 955–65. [Google Scholar]
  101. Truschke, Audrey. 2018. The Power of the Islamic Sword in Narrating the Death of Indian Buddhism. History of Religions 57: 406–35. [Google Scholar] [CrossRef]
  102. Tucci, Giuseppe. 1931. The Sea and Land Travels of a Buddhist Sādhu in the Sixteenth Century. The Indian Historical Quarterly VII: 683–702. [Google Scholar]
  103. Tuladhar-Douglas, Will. 2006. Remaking Buddhism for Medieval Nepal: the fifteenth-century reformation of Newar Buddhism. London: Routledge. [Google Scholar]
  104. Veluppillai, Alvapillai. 2013. Comments on the History of Research on Buddhism among Tamils. In Buddhism among Tamils in pre-colonial Tamilakam and Īlam. Uppsala: Uppsala Universiteit, vol. 3, pp. 58–101. [Google Scholar]
  105. Venkatraman, Ramaswamy. 1990. A History of the Tamil Siddha Cult. Madurai: Ennes Publications. [Google Scholar]
  106. Verardi, Giovanni. 2011. Hardships and Downfall of Buddhism in India. Delhi: Manohar Publishers. [Google Scholar]
  107. White, David Gordon. 1996. The Alchemical Body. Chicago: University of Chicago Press. [Google Scholar]
  108. Yule, Henry, and Burnell Arthur Coke. 1903. A Glossary of Colloquial Anglo-Indian Words and Phrases, and of Kindred Terms, Etymological, Historical, Geographical and Discursive, 2nd ed. London: J. Murray. [Google Scholar]
I use the designation “Nāth” here even though it was not current during the period under consideration (and I do the same for “Hindu”). See (Mallinson 2011, p. 409) for a discussion of the usage of the term Nāth as the name of a grouping of yogi lineages. I use the vernacular form “Nāth” rather than the Sanskrit “Nātha” because it is in vernacular usage that the designation “Nāth” is most usually found.
(Dasgupta 1946, pp. 194–95) denies the possibility of Buddhist origins for the Nāths, partly on the spurious grounds that they were the first alchemists so must have existed before the Pātañjalayogaśāstra because of its mention (4.1) of auṣadhi, medicinal herbs, and thus long predated the esoteric forms of Buddhism with which they have much in common. In east India, Nepal and Tibet, continues Dasgupta, the Nāths’ traditions “got mixed up with those of the Buddhist Siddhācāryas”, a process which was facilitated by their common heritage of tantra and yoga. White (1996, pp. 106–9) suggests that Gorakṣanātha, the second of the human Nāth gurus, was originally Śaiva before being made out in later myths to be Buddhist, concluding that “since no extant tantric or Siddha alchemical works, either Hindu or Buddhist, emerged out of Bengal prior to the thirteenth century, we need not concern ourselves any further with the imagined east Indian Buddhist origins of Gorakhnāth or the Nāth Siddhas”. (White does not address the possibility of elements of Nāth tradition deriving from Buddhist traditions from other parts of India.) Briggs (1938, p. 151 n. 1) names the stages of development of Buddhism in Bengal as Mantrayāna, Vajrayāna and then Kālacakrayāna, and that “[t]hese Buddhist elements were absorbed into the Nāthamārga”. Sen (1956, pp. 281–86) writes of the Nāths’ “Buddhist affiliation” and states that “[b]oth the Nātha cult and Vajrayāna had fundamental unity in their esoteric or yogic aspects”, but makes no suggestion as to which of the two traditions came first. Templeman (1997, p. 957) talks of a “shared praxis” which persisted up to the seventeenth century. In contrast, the polymath writer, poet and scholar M.Govinda Pai, who was from the Tuḷu region (whose Kadri monastery is the focus of this article), says that “the Nātha cult is known to have developed itself out of the Vajrayāna system of the Mahāyāna Buddhism, and thus being in its origin a form of Tāntrik Buddhism before it transformed itself into Tāntrik Śaivism, it naturally betrays no less affinity with the Buddhistic than with the Brāhmanical Tantra” (Pai 1946, p. 64), and his compatriot P. Gururaja Bhatt, who wrote the definitive history of Tuḷunāḍu, concurs (Bhatt 1975, p. 291). Bouillier (2008, pp. 85–86) notes how early historians of the Tuḷu region affirmed that the site was Buddhist before it became Nāth, but that from the 1970s local historians have denied Buddhism’s primacy. Thus Bhatt (1975, pp. 370–72), despite asserting the Nāths’ Buddhist origins, will not entertain the possibility that Kadri was Buddhist before being Nāth and ties himself in various improbable knots trying to defend his position.
(Sanderson 1994, pp. 92–99). See also Sanderson 2001; (Sanderson 2009, pp. 124–20). For summaries of the different positions held in this debate, see (Acri 2015, pp. 263–65) and (Hatley 2016, pp. 37–38).
(Hatley 2016, p. 30), the findings of which article were subsequently refined by Hatley (2018, pp. 114–20) on consideration of tantric medical texts, leading him to conclude that “[f]inding potential intertextuality at the level of the Śaiva Gāruḍa- and Bhūtatantras, Buddhist Kriyātantras, and the early Vidyāpīṭha points toward what is likely to be a history of interaction, shared ritual paradigms, and textual appropriation extending back to the earliest strata of tantric literature”.
Birch 2011, pp. 535–36; Isaacson and Sferra (2014, pp. 100–1); Mallinson (forthcoming b). There is one occurrence of the term haṭhayoga in the c. 3rd century ce Bodhisattvabhūmi (p. 318 ll. 11–17), which is part of the Yogācārābhūmiśāstra.
There have been many Virūpākṣas other than Virūpākṣa the siddha: a Buddhist king known in early Pali sources (Malalasekera 1937 s.v. Virūpakkha; I thank Hartmut Buescher for this reference); one of four great kings of early Mahāyāna in the pre-5th century Suvarṇaprabhāsottama (6.1.1, 6.3.1, 6.6.25; I thank Gergely Hidas for this reference); a form of Rudra mentioned in the Skandapurāṇa (72.64 in the edition in preparation by Peter Bisschop et al.; I thank Professor Bisschop for this reference); one of eight yakṣas in the Śivadharma (pp. 193–208); a form of Śiva whose teachings are given in the c. 12th-century Virūpapañcāśikā; and a form of Śiva which is the central deity of Vijayanagara. Monier-Williams (s.v. virūpacakṣus) gives many more references.
See (Dowman 1985, pp. 43–52) and (Davidson 2005, pp. 49–54) for overviews of Virūpākṣa’s legends (which are first found in the c. 12th-century Grub thob brgyad bcu rtsa bzhi’i lo rgyus of Smon grub shes rab, which is translated in Grunwedel 1916; Robinson 1979 and Dowman 1985); and (Chattopadhyaya 1970, p. 404) for a list of works in Tibetan attributed to him.
A painting from the Drigung tradition which predates 1217 ce is perhaps the earliest Tibetan depiction of Virūpākṣa (Luczanits 2006, p. 82).
In the Grub thob brgyad bcu rtsa bzhi’i lo rgyus, Virūpa is said to have been born at Tripurā in East India and studied at the Somapurī vihāra, which is near Paharpur in Bangladesh (Robinson 1979, pp. 27–28). Tāranātha says that Virūpa lived in Mahrata, i.e., Maharashtra, that he visited Srisailam and that his disciple Kāla Virūpa practised in the Konkan (Templeman 1983, p. 18 and Chattopadhyaya 1970, p. 215). Both Smon grub shes rab and Tāranātha also tell a story of Virūpa destroying an icon of Śiva. The former names the destroyed Śiva as Maheśvara and locates his temple in the unidentified land of Indra (Robinson 1979, pp. 29–30); the latter names the Śiva Viśvanātha and locates his temple in Triliṅga, i.e., the present-day Telangana region (Templeman 1983, p. 15). As a result of a transmission whose details are unknown to me, current Tibetan legend (see e.g., http://www.ludingfoundation.~ org/Archive2016.html accessed 7 June 2018) accords with the c. 1280 ce Marathi Līḷācaritra in locating this episode at Bhīmeśvara, which is one of the three liṅgas referred to in the name of Triliṅga and whose temple complex at Draksharama houses a shrine to Virūpa; on the Līḷācaritra’s story, see footnote 13.
One of the texts attributed to Virūpākṣa in the Vanaratna codex (on which see Isaacson 2008) is entitled Amarasiddhi. Cowell and Eggeling (1876, p. 28) report the name of the text as Amarasiddhiyantrakam but a transcription of the text kindly shared with me by Péter-Dániel Szántó shows that its name is Amarasiddhi (f.47 recto and verso). Chattopadhyaya (1970, p. 404) mentions an Amarasiddhivṛtti among Tibetan works attributed to Virūpākṣa.
Baker 2018, pp. 421–22. Schaeffer (2002, p. 527, n. 12) finds no connection between the Virūpākṣa of the Amṛtasiddhi and the Virūpākṣa of the lam ‘bras tradition, but in the Vanaratna codex described by Isaacson (2008), one of whose texts is, as noted above, an Amarasiddhi of Virūpākṣa, after the text of the Marmopadeśa there is a lineage of teachers which starts from Virūpākṣa and which Isaacson (2008, pp. 3–4) identifies as being very close to some of the lam ’bras lineages.
See e.g., Tucci 1931, p. 690.
Vīra Rājendra Coḍā, who flourished c. 1130 ce, is recorded as having made a donation to Bhīmeśvara at Drākṣārāma in an undated inscription (Epigraphia Indica Vol. IV, p. 51).
I am grateful to Amol Bankar for sharing with me these identifications of Virūpākṣa (personal communication 12 June 2018). Bankar identifies as Virūpākṣa the Dabhoi image, which is one of a group of twelve siddhas of whom some are clearly Nāths (Shah 1957), because, despite considerable damage to the sculpture, it is evident that he is accompanied by a woman and that there are images of the sun and moon above him. Both these motifs are suggestive of the legend of Virūpākṣa in which he stops the sun’s path through the sky so that a lady innkeeper will keep serving him and he will not have to pay his bill (see e.g., Robinson 1979, p. 29). A siddha depicted on the exterior of cave 14 at Panhale Kaji is sitting with a yogapaṭṭa in a posture common in Tibetan images of Virūpākṣa and is accompanied by a woman who may be pouring him a drink. Bankar’s identification as Virūpākṣa of an image of a siddha at Pimpri Dumal (reproduced in Sarde 2014, p. 6, fig. 10) is more tentative, being dependent upon the siddha, who is standing, being accompanied by an anthropomorphic image of Sūrya, the sun god, and pointing at the sky.
I thank Amol Bankar for informing me of the Kāmākhyā Virūpa image, which depicts the tavern episode summarised in footnote 19.
Caryāgīti 3 (Kvaerne 1977, pp. 81–86). I thank Lubomír Ondračka for informing me of this reference.
The middle Indic dohā verses of the Caryāgīti, as well as those of the Dohākośas attributed to the siddhas Saraha and Kāṇha, are usually said to be from east India but this is based on an unwarranted identification of their language as eastern Apabhraṃśa (Szántó forthcoming).
Varṇaratnākara, p. 57.
See footnote 13 and Navanāthacaritramu, p.135 which says that Virūpākṣa was the son of a Maratha king.
The Līḷācaritra was compiled between 1274 and 1287 ce but then lost and reconstructed in the early 14th century (Raeside 1982, p. 491).
Vikramārkacaritramu 6.4 (Jones 2018, p. 199, n. 7).
Tārārahasya, p. 69.
Līḷācaritra pūrvārdha 198–99, Ajñāta Līḷā 27. Elsewhere in the Līḷācaritra there are said to be four oḷīs or lineages of the Nāth tradition, vajroḷī, amaroḷī, siddhoḷī and divyoḷī, of which only the first two are extant in the Kali era (Līḷācaritra, uttarārdh 475). Vajraolī/vajrolī and amaraolī/amarolī are compounds of vajra and amara with olī, which Hemacandra says means kulaparivāṭī, i.e., “lineage” (Deśināmamālā 1.164b; I thank Alexis Sanderson for this reference, which is from the extensive commentary accompanying his forthcoming translation of Abhinavagupta’s Tantrāloka). The distinction between Vajra and Amara lineages, while here being broadly understandable as one between Buddhist and Śaiva traditions, may also reflect a distinction in practice between traditions which engage in sexual rituals (Vajrayāna Buddhists and Kaula Śaivas), and those which spurn such rituals in favour of a method of celibate yoga aimed at jīvanmukti (the Vajrayāna Amṛtasiddhi and the Śaiva Amaraughaprabodha). Ultimately the celibate yoga tradition of the Amaraughaprabodha came to dominate the Nāth saṃpradāya.
I thank Amol Bankar, who is editing the Tattvasāra, for providing me with this information.
Navanāthacaritramu, p. 135.
This part of the story is perhaps an echo of an episode in Tibetan treatments of Virūpākṣa in which he is rebuked by his fellow monks for eating pigeons, prompting him to abandon his monastery and then restore the pigeons to life (see e.g., Dowman 1985, pp. 44–46).
His disciples are named as Rasendrapāya, Ratnapāya, Uccaya, Kālapāya, Vajrakākanātha, Jālāndhra, Śaindrapāla (Jālāndhra’s student), Kāmaṇḍa, Pūrṇagirinātha, Endiyāṇiguru, Bhuvanendra and Trilocanasiddha (Navanāthacaritramu, pp. 211–14).
Vikramārkacaritramu 6.4 (Jones 2018, p. 199, n. 7).
Haṭhapradīpikā 1.9.
Both mīna and matsya mean fish. In Śaiva texts which predate the haṭha corpus, Mīnanātha and Matsyendra are one and the same (see e.g., Devīdvyardhaśatikā vv. 161c-163b and Ciñcinīmatasārasamuccaya 7.50 (I thank Alexis Sanderson for these references), in both of which Mīnanātha is given as the name of the propagator of the Kaula Pūrvāmnāya, who elsewhere is identified as Macchanda (Tantrāloka 1.7), Matsyendra (Kaulajñānanirṇaya 11.43) and Macchaghna (Kaulajñānanirṇaya chapter colophons)). Subsequent Tibetan and Indian siddha lists and depictions include two fish-related siddhas (e.g., Lūyipa and Mīnapa in the Grub thob brgyad cu rtsa bzhi’i lo rgyus, Matsyendra and Mīna in the Haṭhapradīpikā (1.5), and statuary in Maharashtra (Sarde 2017)).
Mīnanātha is invoked in the maṅgala verse of the long recension of the Amaraughaprabodha. The hemistich in which his name is given (after Ādinātha) is not found in the two manuscripts of the older, short recension, but is likely to have dropped out in the course of their transmission. Two siddhas are mentioned in the other hemistich, Cauraṅgi and Siddhabuddha. Their close association in legend with Matsyendra indicates that Mīnanātha here is another name for Matsyendra.
Kadalīmañjunāthamāhātmya chps. 48–53.
See footnote 13.
See footnote 19.
Virūpākṣa is included, together with Mañjunātha, in a list of 84 siddhas current at the Nāth headquarters in Gorakhpur: This is likely to be due to their inclusion in the list of nine Nāths in the Kadalīmañjunāthamāhātmya, which was edited under the auspices of the Nāth order in 1956.
Exceptions are tantric Buddhist images from the Himalayan region, the Caryāgīti, whose place of composition is uncertain, and the Maithili Varṇaratnākara.
Navanāthacaritramu 213 (the title is given as Amṛtajñasiddhi). By identifying Virūpākṣa as the author of the Amṛtasiddhi and locating his origins in present-day Maharashtra, the Navanāthacaritramu points to that region as the place of composition of the Amṛtasiddhi, an inference supported by parallels between the Amṛtasiddhi and the c. 12th- or 13th-century old Marathi Vivekadarpaṇa, whose teachings on the yogic body are similar to those of the Amṛtasiddhi (see e.g., Vivekadarpaṇa ch. 5 on the sun and moon) and which, like the Amṛtasiddhi, gives its chapters the unusual designations of viveka and lakṣaṇa. On the date of the Vivekadarpaṇa, see Reinelt 2000, pp. 93–95.
Navanāthacaritramu canto 1 (Jones 2018, pp. 185–86).
Navanāthacaritramu cantos 4–5 (Jones 2018, pp. 189–94).
The Kadalīmañjunāthamāhātmya is a palimpsest of Buddhist, Śaiva and Vaiṣṇava teachings, reflecting the passing of the control of the Mañjunātha temple from Vajrayāna Buddhists to Nāth Śaivas and then Mādhva Vaiṣṇavas, with the latter responsible for its final redaction. For an overview of its contexts and context, see Nagaraju 1969.
Kadalīmañjunāthamāhātmya 14.9, 48.7, 50.11, 50.17, 53.27. The c. 15th-century Ānandakanda, which was composed at Srisailam, gives a list of nine Nāths whose fourth is Koṅkaṇeśvara and may correspond to Virūpākṣa (1.3.47a-48b):
  • ādināthaṃ mīnanāthaṃ gorakṣaṃ koṅkaṇeśvaram |
  • jālandhreśaṃ kandhanīśam oḍḍīśaṃ ciñciṇīśvaram ||
  • cauraṅgim etān nāthākhyān nava saṃtarpayet tataḥ |
See (Schoterman 1975, pp. 934–35); (Sanderson 2011, pp. 44–45 and 2014, pp. 62–64), and (Mallinson 2011, pp. 412–14). The Śaiva tradition of the southern Nāths is Śāmbhava, a variant of the Paścimāmnāya (Kiss forthcoming). The oldest known statues of Nāths date from the 12th century ce onwards and are found in western India and the Deccan (Sarde 2017, pp. 96, 108–10).
Tantrāloka 29.32.
Kiss (forthcoming, p. 32) says that the Matsyendrasaṃhitā was “composed in South India, probably in the Tamil region, or alternatively, around Goa, in the 13th- century”. Kiss chooses the Tamil region over the Konkan because of the mention in the Matsyendrasaṃhitā of the god Śāstṛ, whom he identifies as a specifically Tamil deity, but the Kadri Mañjunāth temple has an image of Śāstṛ dated to the twelfth century and other, earlier images of Śāstṛ are found in the region (Bhatt 1975, pp. 354–55) and plates 290 and 291. Of the two toponyms mentioned in the Matsyendrasaṃhitā, the first, Gomanta, is likely to be in present-day Goa (Kiss forthcoming, pp. 30–31) while the location of the second, Alūra, whose king’s dead body was taken over by Matsyendra, is uncertain. Kiss gives various possible identifications of Alūra, including Ellora, Eluru in Andhra Pradesh and Vellore (Kiss forthcoming, p. 31). The five manuscripts of the Matsyendrasaṃhitā all have the reading alūrādhipatiṃ bhūpaṃ dākṣiṇātyaṃ purā kila (55.3ab). The text is likely to have been redacted in the early 19th century at the request of the Mahārāja of Jodhpur (Kiss forthcoming, p. 32); perhaps the original reading, obscure to north Indian scribes, was ālupādhipatiṃ, referring to a king of the Ālupa dynasty, which ruled the Tuḷu region from the early centuries ce until the end of the 14th century (Bhatt 1975, p. 18), and whose capital was at Mangalore from the 11th to 13th centuries, the likely period of composition of the Matsyendrasaṃhitā. In the Navanāthacaritramu, the king who plays the same rôle in the story of Matsyendra rules over Kadri, which at the time would have been under Ālupa rule. The Matsyendrasaṃhitā does not, however, mention Kadri in its version of the story of Gorakṣa rescuing Matsyendra. Supporting a Tamil origin for the Matsyendrasaṃhitā is its identification of Gorakṣa as a Coḷā king (paṭalas 1 and 55) and its naming at 55.20 of Matsyendra’s son and grandson as Kharparīśa and Vyālīndra, both of whom are associated with alchemy in various Tamil traditions but are not mentioned in texts associated with Kadri (or Srisailam).
Deshpande 1986. Panhale Kaji has various Nāth statues, including a group of 9 or perhaps 10 siddhas. A group of 12 Nāth siddhas, dating to 1230 ce, is found at Dabhoi near Ahmedabad in Gujarat. A statue of Matsyendra from the Kadri monastery and now in the Mangalore Government Museum has been dated to the 10th century, but, as will be explained below, this date is likely to be too early and was probably proposed for ideological reasons.
Bhatt 1975, p. 299 and plates 303 and 304(a).
I thank Manu Devadevan for sharing with me his transcription and translation of this inscription. Bouillier (2008, p. 96) reports that the inscription refers to Candranātha as a king (arasu) but Devadevan tells me that this is not the case. Bhatt (1975, p. 295) notes that an inscription from Mangalore dated 1434 ce mentions “the gift of land to one Jugādikuṇḍala Jōgi-Puruṣa by Jōgi-Oḍeya alias Chauṭa”.
Bhatt 1975, p. 294.
Badger 1863, pp. 111–13.
Grey 1892, pp. 345–52.
Tashrīh al-Aqvāṃ chp. 104 (British Library Board, Add.27255 f.399). I thank Bruce Wannell for translating this passage for me.
This migration is recorded in the Teyyam performances regularly put on by the Cōyī (the vernacular for yogī) caste in northern Kerala, in which it is precipitated by the death of the king’s son and the resultant ending of his lineage (Freeman 2006, pp. 167–69).
On the rājyābhiṣeka, see Bouillier 2008, chp. 6.
(Sastri 1939, pp. 104–5). On early traces of Buddhism in Karnataka see (Nagaraju 1983, pp. 6–10).
Acri 2016, p. 8.
(Szántó 2016b, p. 2). The exact location of Mahābimba is uncertain. I know of two further possible references to it. Tāranātha says that in Koṅkaṇa his guru Buddhaguptanātha saw “the self-created image of Mañjuśrī in the middle of a pond. It is called Jñānakāya... Then he saw also the bimbakāya which looks like a rainbow raising the stūpa of the accumulated vapour beyond touch” (Tucci 1931, p. 696). A manuscript of the Aṣṭasāhasrikā Prajñāpāramitā dated 1015 ce mentions a pilgrimage site in Koṅkaṇa called Mahāviśva, which Szántó (2012, Vol. 1/2 p. 40, fn. 61) suggests may be a corruption of Mahābimba. A puzzling verse in the long recension of the Amaraughaprabodha (67 in Jason Birch’s edition, from which the variant readings below are taken), which does not appear to fit its context and is also found, with significant variants, at Sekoddeśa 26, indicates that these two unusual compounds may refer to a single object, which lights up various heavenly bodies (including smoke (dhūma ) and specks of dust ( marīci ), suggestive of the vapour reported by Tāranātha):
  • dhūmamarīcikhadyotadīpajvālendubhāskarāḥ |
  • amūḥ kalā mahābimbaṃ mahāviśvaṃ prakāśate ||
  • a dhūma ] T 1 Sekoddeśa (contra metrum); rumo B 1 A 1 G 2 Ed
  • b bhāskarāḥ ] bhāskaraiḥ Sekoddeśa
  • c amūḥ ] em. Mallinson; amī codd., tamaḥ Sekoddeśa • mahābimbaṃ ] mahābindur Sekoddeśa
  • d mahāviśvaṃ prakāśate ] viśvabimbaṃ prabhāsavaram Sekoddeśa
Bhatt 1975, plate 298(a). A relief of the Buddha still in situ at Kadri has been dated to the ninth or tenth centuries (Bhatt 1975, plate 304(b)).
Szántó 2015, pp. 546 and 550–52.
Szántó 2015, p. 558.
Davidson 2002, p. 159.
Hevajratantra 1.6.16.
Cambridge University Library Add. 1643.
These are Lokanātha in Śrīkhairavaṇa, Sahasrabhuja Lokanātha in Śivapura, Lokanātha at Mahāviśva, Khaḍgacaitya in Kṛṣṇagiri (i.e., Kanheri), a caitya on Pratyekabuddha peak at Kṛṣṇagiri and a Lokanātha caitya at Marṇṇava (Kim 2014, pp. 42, 44, 66–68).
Epigraphia Carnatica VII Sk 170.
Epigraphia Carnatica VII Sk 169.
(Nagaraju 1983, pp. 12–13 and fig. 9), which draws on the 1941 Annual Report of the Mysore Archaeological Department but has a better picture of the statue.
Bhatt 1975, p. 373 and plate 304a (b).
Indian Antiquary X, pp. 185ff.; Annual Report on South Indian Epigraphy 1927–28, Appendix E, Nos. 65–66.
Desai 1954, p. 91.
Bhatt 1975, pp. 372–73.
See (Templeman 1983, pp. 17–18) on Virūpa’s disciple Kāla Virūpa practising asceticism in the Konkan before returning to his guru in the Maratha region; (Templeman 1983, p. 81) on the siddha Nāgopa; (Templeman 1995, p. 5) on king Haribhadra, who mastered the siddhi of making magical pills; and (Chattopadhyaya 1970, p. 351) on Jñānākaragupta.
Templeman 1983, pp. 82–83.
(Yule and Burnell 1903, p. 28) s.v. Anchediva:
“c. 1345.—Ibn Batuta gives no name, but Anjediva is certainly the island of which he thus speaks: “We left behind the island (of Sindābūr or Goa) passing close to it, and cast anchor by a small island near the mainland, where there was a temple, with a grove and a reservoir of water. When we had landed on this little island we found there a Jogi leaning against the wall of a Budhkhānah or house of idols.”
Ibn Batuta, iv. 63.”
Anjediva was taken over by the Portuguese in 1505 (Mathew 1988, p. 163). The name of the island given by Tāranātha appears to be San Lorenzo, but the Portuguese church on Anjediva has always been known as Nossa Senhora das Brotas (Our Lady of the Springs) in homage to the island’s good supply of fresh water. San Lorenzo was a name for Madagascar, but that is by no means a “small island”, so it seems that either Tāranātha was again conflating his sources or he was, as suggested by Templeman (1997, p. 962, n. 38), referring to a Portuguese settlement by that name elsewhere in the Konkan. The only such reference I have found is to a church of San Lorenzo in Goa mentioned by della Valle in the early 17th century, but which was no longer standing in the 19th century (Grey 1892, p. 495).
Anjediva is now under the control of the Indian Navy and closed to visitors, including local Christians wanting to visit its two churches. I visited the neighbouring Kurumgad, another possible candidate for Tāranātha’s island, in March 2016, only to discover that it was covered in Portuguese fortifications dating to the beginning of the 16th century and to be told that all the other habitable islands in the vicinity were similarly fortified.
Tucci 1931, p. 692, n. 2.
The location of a thousand-armed Avalokiteśvara at a site called Śivapura in a 1025 ce manuscript of the Aṣṭasāhasrikā Prajñāpāramitā (see p. 10) indicates another possible site of Vajrayāna-Śaiva interaction (I thank Andrea Acri for suggesting this in an email dated 24th January 2019). Bankar (2013) and Sarde (2016) describe early Buddhist sites in the Deccan which were later occupied by Nāths, but there is no evidence of Vajrayāna Buddhism at the sites, nor of direct links between the Buddhist and Nāth traditions.
Deshpande (1986, p. 121) dates the oldest caves at Panhale Kaji to before the 5th century, but Rees (forthcoming) has shown that a later date, probably in the latter half of the first millennium, is more likely.
See figure
Deshpande 1986, pp. 146–48.
On Kadri, see Bouillier 2008, chp. 4 and Bouillier 2009; also Saletore 1937; Pai 1946; (Bhatt 1975, pp. 287–97) and (Freeman 2006, pp. 164–67).
See South Indian Inscriptions Vol. VII No. 191, in which the transcription of the part of the inscription which gives its year reads thus:
  • kalau va-[r]-ṣa-sahasrāṇām-atikrāntē catuṣṭaye |
  • pu-[narabda]-gatē caivāṣṭaṣaṣṭyā samanvitē
As it stands this is an unlikely formulation which must correspond to the year 4069 (4000 + 68 + 1) of Kaliyuga, i.e., 967–68 ce, and this date has been repeated in most secondary literature on Kadri. Pai, however, without viewing the inscription itself, demonstrated that gate must be a misreading of śate (ga and śa are similar in the Grantha script in which the inscription is written), not only because śate is much better Sanskrit, but also because the tithi given only makes sense if the year is 1068 rather than 968 (Pai 1946, pp. 60–62). Dominic Goodall visited Kadri in 2017 and confirmed that the correct reading is śate (personal communication 6th February 2017).
Kadalīmañjunāthamāhātmya 13.10–11:
  • uttarārāmamadhye tu sthito viṣṇur manoharaḥ |
  • ālaye mañjunāthathasya pūrvottaragato hariḥ ||
  • pūrvadakṣiṇayor bhāge bhagavāṃś caturānanaḥ |
  • tayor uttarabhāge tu vyāsaḥ satyavatīsutaḥ ||
“In the middle of the northern grove is situated a delightful Viṣṇu. In the abode of Mañjunātha, to the northeast, is Hari. In the area to the southeast is the four-faced Lord [i.e., Brahmā]. To the north of both of them is Vyāsa, the son of Satyavatī.”
I do not know what image is being referred to as Viṣṇu here; it may no longer be at the site.
Alexis Sanderson first suggested this to me, in a meeting in December 2010. His identification was subsequently confirmed to me by Christian Luczanits (email communication 30 January 2018).
I thank Christian Luczanits for this identification (email communication 26 April 2016).
I thank Christian Luczanits (email communication 26 April 2016) and Andrea Acri (email communication 24 January 2019) for these observations.
Comparison of photographs of the image show that its composition has been altered in recent years, and confirm that the mūrti is not of one piece with the plinth. The photographs in the Government of Madras’s Annual Report for Epigraphy for the Year Ending 31 March 1921 (Plate 1), Bhatt (1975, plates 300 and 301) and Shetti (1988, fig. 10) show the prabhāvalī (halo) behind Mañjuvajra higher than it is today, and the image now has a metal sheet between it and the plinth. The sheet has small posts at its rear which support the prabhāvalī in its new position. In the oldest photograph Mañjuvajra holds a lotus flower in each of his middle left and right hands as he does today, while in the photographs found in Bhatt 1975 and Shetti 1988 the lotuses are not in Mañjuvajra’s hands but in those of his two attendants.
The image is identifiable as Matsyendra (whose name means “lord of fish”) because the subject is seated on a fish.
The Lokeśvara image has four arms and his other right hand may have held something, but is now empty. Matsyendra, being human-born, has only two arms.
Grey 1892, p. 348.
Personal communication from Shreekant Jadhav 27th February 2019.
South Indian Inscriptions Vol. VII No. 189 (pp.84–85).
The head or Rājā of the monastery does play a role in the temple’s annual car festival (Nagaraju 1969, p. 68).
Photographs of the statues without their silver coverings are reproduced by Bhatt (1975, plates 303 and 304(a)). Cauraṅginātha (“Chowranginatha” on the label) is referred to as Śṛṅginātha by Bhatt (1975, p. 299). As noted by Bouillier (2008, p. 84), no source is known for the ascription of the names to these images. They now have full-length silver covers replicating, with embellishment, the iconography of the stone images, but at the base of the cover on the image said to be of Matsyendra is a fish which is not found on the image beneath.
Bhatt 1975, p. 299 and plates 303 and 304(a).
Bhatt 1975, p. 290, n. 1, p. 296, n. 27. The best known temple of Śiva Mañjunātha is at Dharmasthala, c. 50 kilometres inland from Mangalore. The Mañjunātha liṅga at Dharmasthala, which is of the usual iconic form unlike the simple svayambhū liṅga at Kadri, is said to be the original object of worship at Kadri Mañjunātha and to have been transported to Dharmasthala from Kadri in the 16th century by Vādirāja, the maṭhādhipati of the Mādhva maṭha at Udipi (Nagaraju 1969, p. 68).
Kandahjaya 2016, p. 88 and, n. 60, table p.91. The Vimalaprabhā commentary on the Kālacakratantra, whose teachings have some parallels with those of the Amṛtasiddhi, opens with an invocation to Mañjunātha.
See page 23.
Loc. cit.
South Indian Inscriptions Vol. VII No. 189 (pp. 84–85) dated śaka 1308 prescribes materials for the worship of Śrīmañjinātha of Kadaḷi. No.190 (pp. 85–87) dated śaka 1311 is very hard to read, but mentions Mañjinātha and possibly the establishment of an image at the temple. I thank Manu Devadevan for sharing with me his transcriptions and translations of these Kannada inscriptions.
Grey 1892, p. 348.
Bouillier 2008, p. 96 and n. 42, citing a transcription by Ānandanāth Jogī.
The actual name Mangalore is not given; Gauraṇa writes of “a large city on the western shore rich in auspiciousness (maṅgalāvṛtam)”: poṃgāru paścimāṃbudhitīramunanu maṃgaḷāvṛtam agu mahanīyam aina puṭabhedanamu (Navanāthacaritramu, p.176).
The 55th and last paṭala of the Matsyendrasaṃhitā, which dates to perhaps the 13th century, tells a version of the story similar to that found in the Navanāthacaritramu (see Kiss forthcoming, p. 12 for a synopsis). Ondračka (2011) analyses medieval Bengali versions of the legend and Muñoz (2011, pp. 115–27) gives summaries and analysis of current north Indian versions. The Kaulajñānanirṇaya, which is ascribed to Matsyendra and is transmitted in a manuscript dated to the mid-11th-century on palaeographic grounds (nak ms. 3–362/ngmpp A48/13; I thank Shaman Hatley for the dating (personal communication 15th January 2018)), tells the story of Matsyendra overhearing Śiva teach Pārvatī the Kaula doctrine but not that of his sojourn at Kadri, suggesting that it predates the events on which the legend is likely to be based.
This is perhaps the Narendra Hill in Sawantwadi, southern Maharashtra. I thank Jason Birch for this suggestion.
Suprabhedāgama yogapāda nāḍīcakravidhipaṭalaḥ 3.6–7, Sahasrāgama kriyāpāda 45.6–7, Yogajāgama pavitrārohaṇapaṭalaḥ verse 58 and mahotsavapaṭalaḥ verse 166. Cf. Goodall (2011, p. 224) on the kūrmaśilā, the turtle-stone which supports a liṅga.
Navanāthacaritramu, p.94.
I thank Jamal Jones for this observation (personal communication 23 March 2018).
(Sanderson 2005, pp. 135–36, n. 345) writes “[t]he term *Anuttarayogatantra that has long been current in academic writing on Tibetan Buddhism does not occur to my knowledge in any Sanskrit text but is an erroneous reconstruction from Tibetan rnal ’byor bla med kyi rgyud, which rather renders the Sanskrit yoganiruttaratantram”. It is noteworthy that unlike the Haṭhapradīpikā, which teaches royal (rāja) yoga and haṭha yoga, the Navanāthacaritramu uses the name anuturya rather than haṭha or a related term, perhaps because of haṭha’s negative connotations (on which see Mallinson forthcoming b).
See the forthcoming edition of the Amṛtasiddhi by Péter-Dániel Szántó and me for details.
Szántó 2015, p. 543.
See (Birch 2011, p. 535) and (Mallinson forthcoming b).
Amaraughaprabodha (short recension) verse 7.
p. 13.
On Mīnanātha and Matsyendranātha, see note 35.
The first half of the maṅgala verse, which mentions Ādinātha and Mīnanātha, is absent in the two manuscripts of the older recension of the Amaraughaprabodha.
Cauraṅgi does not occur in northern lists of the nine Nāths, but he is included in a list of 84 siddhas given in the early 14th-century Maithili Varṇaratnākara (p. 57), in which his name comes third, after Mīnanātha and Gorakṣaṇātha [sic]. His story was current in Bengal and Mithila (it is told in the Gorakh Vijay cycle) and is popular in Punjab, where he is known as Pūraṇ Bhagat (see e.g., White 1996, pp. 298–99). Cauraṅgi’s legend is also found in the Grub thob brgyad cu rtsa bzhi’i lo rgyus, a 12th-century Tibetan account of the lives of the 84 siddhas (the story is similar to that taught in the Navanāthacaritramu) and he is mentioned in the 13th-century Marathi Jñāneśvarī (pp. 1730–40).
The bulk of the Kadalīmañjunāthamāhātmya, chp. 15–45, is devoted to their exploits.
Haṭhapradīpikā 1.6.
Jones 2018, p. 199, n. 7: siddhabuddhuni buddhicittaṃbunaṃ jerci.
Similarly, the Nāth presence at Panhale Kaji did not result in the destruction or removal of Buddhist statuary.
Thus Amaraughaprabodha (short recension) 35 is a reworking of Amṛtasiddhi 19.15, in which the Vajrayāna concept of the vicitrakṣaṇa, one of four kṣaṇas or “moments” associated with the four blisses experienced during sexual ritual, becomes vicitrakvaṇaka, “having variegated tones”, an epithet of the anāhata, “unstruck”, sound heard internally by the yogi. At vv. 38 and 40 of its short recension the Amaraughaprabodha retains the Amṛtasiddhi’s teachings on the Vajrayāna concepts of paramānanda and sahajānanda, and similarly, at vv. 39 and 40, those of atiśūnya and mahāśūnya.
See Bajracharya and Michaels 2016, pp. 5–51 for a translation of a legend explaining the identification of Avalokiteśvara and Matsyendra in the 1830 ce Nepālikabhūpavaṃśāvalī. Of particular note is that Lokeśvara became Matsyendra after overhearing the Kaulajñāna, the Kaula doctrine which is the subject of the Kaulajñānanirṇaya of Matsyendra (on which see footnote 117). See also Locke 1980, pp. 288–89 and chp. 13, and Tuladhar-Douglas 2006, pp. 181–82.
Bajracharya and Michaels 2016, pp. 69–72, Malla 2006, pp. 4–9. Cf. Michaels 1985 and Lévi 1905, pp. 218–21, 353 (in the latter reference, Lévi names Bengal, Malabar and Sri Lanka as possible sources for the Nepalese Matsyendra cult).
The 15th-century Guṇakāraṇḍavyūha identifes Avalokiteśvara and Matsyendra (Tuladhar-Douglas 2006, p. 7). Arguing against claims of an earlier identification is the absence of mentions of Matsyendra, Gorakṣa or Lokeśvara in the late 14th-century Gopālarājavaṃśāvalī.
On the former, see Verardi 2011; on the latter as a recurrent historical trope, see Truschke 2018.
Desai 1954, pp. 91–92.
See footnote 13.
On which see p. 13.
Līḷācaritra pūrvārdha 198–99, A.27. I thank Amol Bankar for this reference and for sharing with me his translation of the relevant passage.
I thank Alexis Sanderson for this reference, which is from a draft of his forthcoming commentary on the Tantrāloka.
Ṣaṭsāhasrasaṃhitā ff. 351v-353r, for an edition of which by Alexis Sanderson and me, see https:\protect\discretionary{\char\hyphenchar\font}{}{}//www.\protect\discretionary{\char\hyphenchar\font}{}{}academia.\protect\discretionary{\char\hyphenchar\font}{}{}edu/35397806/Mallinson$_$Harvard$_$Talk$_$Nov$_$2017_Handout. The instruction takes place at Arbuda, i.e., Mount Abu. There are no material remains at Abu from the period under consideration which indicate either a Buddhist or Nāth presence, but there is material evidence for both traditions at Taranga, a site 70 kilometres due south of Abu best known for its magnificent 13th-century Jain Ajītnāth temple. On the opposite side of the Siddhaśilā hill at Taranga from the Ajītnāth temple is a small temple complex with two shrines containing images now worshipped as Hindu goddesses called Dhāraṇ Mātā and Tāraṇ Mātā. Dhāraṇ Mātā is flanked by six slightly smaller sculptures. The priests at the shrine dress her in the attire of a goddess, but she and the others are all male figures whose iconography indicates that they are Buddhist, with some dating to perhaps the 6th century (I thank Ken Ishikawa for this tentative dating). Tāraṇ Mātā is a beautiful female image who has been identified as Kurukullā (Sompura 1969, pp. 29–30); see also (Sompura 1968, p. 72) and (Rawat 2009), but is in fact the Buddhist goddess Tārā, as can be inferred from her retinue (I thank Christian Luczanits for this identification (personal communication 21st June 2017)). On the hill itself is a cave, known locally as jogan dā guphā, “the cave of the yogis”, which has a small makeshift shrine centred on a freestanding panel depicting in relief four Buddhist images of seated meditating figures. On the top of the hill is a small temple from perhaps the 19th-century (I thank Crispin Branfoot for confirming my tentative dating), which now houses the pādukā of a Jain saint but has eight statues of Nāth yogis on its roof, suggesting that it was originally a Nāth shrine. Tāranātha mentions a yoginī disciple of Śāntigupta (his guru’s guru) called Tāraṅgā, who travelled widely, including to the Konkan (Templeman 1983, p. 93).
E.g. Matsyendra, who is mentioned in the early 11th-century Tantrāloka as the propounder of the Śaiva Kaula doctrine in the kali age (1.7, 29.32 and 30.102), or Virūpa, who is Buddhist in early sources such as the Grub thob brgyad cu rtsa bzhi’i lo rgyus and the Līḷācaritra.
E.g. Gorakṣa, whose earliest textual mentions, which date to the 13th century, are found in both Śaiva and Buddhist texts (Mallinson 2014, p. 233, n. 28).
Thus for Schaeffer, the first scholar to study closely the Amṛtasiddhi, it “cannot be comfortably classified as either Buddhist or non-Buddhist” (2002, p. 515).
See (Szántó 2017, pp. 7–8) on the Gāhaḍāvala Jayacandra, and the Nityakaumudī (ngmpp b35–26 and nak 4/324), which is a commentary on the Śaiva Vīracandra’s 1072 ce Nityaprakāśa whose colophon says that it was commissioned by the Pāla monarch Rāmapāla’s guru, Śambhudatta. I thank Péter-Dániel Szántó for this reference.
(Gokhale 1990, p. 70); see ibid., p. 10 for more on connections between Kanheri and Gauḍa.
See footnote 50.
Matsyendrasaṃhitā paṭalas 1 and 55.
The Tirumantiram mentions the Kālacakra[tantra] (Tirumantiram “III chp.14”; see (Venkatraman 1990, p. 193) for this reference, which I have been unable to confirm). The Kālacakratantra may be dated to between 1025 and 1040 ce (Newman 1998). The Tirumantiram is cited in a commentary on the Yāpparuṅgalakkārigai by Guṇasāgara, who was active c. 1100 (Venkatraman 1990, p. 193). A text-critical study of the Tirumantiram remains a desideratum; if the text as it is currently constituted does not contain additions to its earliest layer, it may thus be dated to the second half of the 11th century ce. (On the implausibility of the very early datings of the Tirumantiram often found in secondary sources, see Goodall 1998, p. xxxvii, n. 85 and Goodall 2004, pp. xxix-xxx.)
The Tirumantiram makes no mention of the Nāth tradition but its doctrinal parallels with the Matsyendrasaṃhitā lead Kiss (forthcoming, pp. 51–52) to conclude that the two texts are “not completely unrelated”. The Tirumantiram’s yoga method has much in common with that of the Amṛtasiddhi, but, like the haṭhayoga of other Vajrayāna texts and in contrast with the celibate yoga of the Amṛtasiddhi, includes sexual intercourse without ejaculation. (In the sexual ritual taught in paṭala 40 of the Matsyendrasaṃhitā the yogi is to ejaculate.) The Tirumantiram mentions the Buddhist Kālacakratantra (see note 156) and celebrates the Śaiva site of Chidambaram, which is also a cultic centre for the Keralan caste yogis who moved south from Kadri in perhaps the 17th century (Freeman 2006, pp. 172–73).
The Kadri Lokeśvara, for example, resembles a Coḷā Śiva sold by Christies in New York in 2015:
Kandahjaya (2016, p. 160) notes that the Śāriputra has been variously dated to between the 5th to 15th centuries. Its mention of Matsyendra indicates that it was composed in the latter few centuries of this period. Paranavitana (1928, pp. 60–62) edits the verses describing Avalokiteśvara thus:
  • caturbhujaṃ trinetraṃ ca pāśadaṇḍāyudhaṃ tathā |
  • kuṇḍikāhomahastaṃ ca purībhadrasamaulikam ||
  • śaṅkhakambalahāraṃ ca raktāmbaradharānvitam |
  • yugamatsyasamārūḍhaṃ matsyendranātham eva ca ||
Avalokiteśvara/Nātha is identified with eight Nāthas in the Śāriputra: Śiva Nātha, Brahma Nātha, Viṣṇunātha, Gaurī Nātha, Matsyendra Nātha, Bhadra Nātha, Bauddha Nātha and Gaṇa Nātha (ibid.). Additionally, Avalokiteśvara’s role as the protector of mariners in Sri Lanka (Bopearachchi 2014, p. 182), is echoed by Matsyendra’s identification as a fisherman in many of the various legends associated with him.
Jayabhadra gives his place of birth in his (Szántó 2016a, p. 2).
See (Scheurleer 2008, p. 292, n. 21). I thank Andrea Acri for drawing my attention to this statue (personal communication 24th January 2019).
(Veluppillai 2013, pp. 65–77); see also (Acri 2018, p. 13).
Saletore (1936, pp. 384–85) argues that the strong Śaivism of Tuḷu rulers was responsible for disappearance of Buddhism from the Tuḷu region.
Jaini 1980; see also (Bouillier 2008, p. 112).
See (Bisschop 2018, p. 396) on Avalokiteśvara as a Buddhist Īśvara in the pre-630 ce Kāraṇḍavyūhasūtra.
Figure 1. Virūpākṣa, Mahudi Gate, Dabhoi.
Figure 1. Virūpākṣa, Mahudi Gate, Dabhoi.
Religions 10 00273 g001
Figure 2. Virūpākṣa, Panhale Kaji cave 14.
Figure 2. Virūpākṣa, Panhale Kaji cave 14.
Religions 10 00273 g002
Figure 3. Acala, Panhale Kaji cave 10.
Figure 3. Acala, Panhale Kaji cave 10.
Religions 10 00273 g003
Figure 4. Mañjuvajra, Mañjunātha temple, Kadri.
Figure 4. Mañjuvajra, Mañjunātha temple, Kadri.
Religions 10 00273 g004
Figure 5. Lokeśvara, Mañjunātha temple, Kadri.
Figure 5. Lokeśvara, Mañjunātha temple, Kadri.
Religions 10 00273 g005
Figure 6. Buddha, Mañjunātha temple, Kadri.
Figure 6. Buddha, Mañjunātha temple, Kadri.
Religions 10 00273 g006
Figure 7. Matsyendra, Mangalore Museum.
Figure 7. Matsyendra, Mangalore Museum.
Religions 10 00273 g007
Figure 8. Kadri Monastery, Matsyendra.
Figure 8. Kadri Monastery, Matsyendra.
Religions 10 00273 g008
Figure 9. Matsyendra, Mangalore Museum (detail).
Figure 9. Matsyendra, Mangalore Museum (detail).
Religions 10 00273 g009
Figure 10. Matsyendra, Kadri Monastery (detail). Photograph by Jacqueline Hargreaves.
Figure 10. Matsyendra, Kadri Monastery (detail). Photograph by Jacqueline Hargreaves.
Religions 10 00273 g010
Figure 11. Lokeśvara (Amitābha detail), Mañjunātha Temple.
Figure 11. Lokeśvara (Amitābha detail), Mañjunātha Temple.
Religions 10 00273 g011
Figure 12. Mañjuvajra (Akṣobhya detail), Mañjunātha Temple.
Figure 12. Mañjuvajra (Akṣobhya detail), Mañjunātha Temple.
Religions 10 00273 g012
Figure 13. Matsyendra, Mañjunātha temple, Kadri.
Figure 13. Matsyendra, Mañjunātha temple, Kadri.
Religions 10 00273 g013
Figure 14. Cauraṅgi, Mañjunātha temple, Kadri.
Figure 14. Cauraṅgi, Mañjunātha temple, Kadri.
Religions 10 00273 g014
Figure 15. Goraksa, Mañjunātha temple, Kadri.
Figure 15. Goraksa, Mañjunātha temple, Kadri.
Religions 10 00273 g015

© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (
Religions EISSN 2077-1444 Published by MDPI AG, Basel, Switzerland RSS E-Mail Table of Contents Alert Logo copyright Steve Bridenbaugh/UUA
Back to Top